>> From theCUBE Studios in Palo Alto and Boston, bringing you data-driven insights from theCUBE and ETR. This is breaking analysis with Dave Vellante. >> Moore's Law is dead, right? Think again. Massive improvements in processing power combined with data and AI will completely change the way we think about designing hardware, writing software and applying technology to businesses. Every industry will be disrupted. You hear that all the time. Well, it's absolutely true and we're going to explain why and what it all means. Hello everyone, and welcome to this week's Wikibon Cube Insights powered by ETR. In this breaking analysis, we're going to unveil some new data that suggests we're entering a new era of innovation that will be powered by cheap processing capabilities that AI will exploit. We'll also tell you where the new bottlenecks will emerge and what this means for system architectures and industry transformations in the coming decade. Moore's Law is dead, you say? We must have heard that hundreds, if not, thousands of times in the past decade. EE Times has written about it, MIT Technology Review, CNET, and even industry associations that have lived by Moore's Law. But our friend Patrick Moorhead got it right when he said, "Moore's Law, by the strictest definition of doubling chip densities every two years, isn't happening anymore." And you know what, that's true. He's absolutely correct. And he couched that statement by saying by the strict definition. And he did that for a reason, because he's smart enough to know that the chip industry are masters at doing work arounds. Here's proof that the death of Moore's Law by its strictest definition is largely irrelevant. My colleague, David Foyer and I were hard at work this week and here's the result. The fact is that the historical outcome of Moore's Law is actually accelerating and in quite dramatically. This graphic digs into the progression of Apple's SoC, system on chip developments from the A9 and culminating with the A14, 15 nanometer bionic system on a chip. The vertical axis shows operations per second and the horizontal axis shows time for three processor types. The CPU which we measure here in terahertz, that's the blue line which you can't even hardly see, the GPU which is the orange that's measured in trillions of floating point operations per second and then the NPU, the neural processing unit and that's measured in trillions of operations per second which is that exploding gray area. Now, historically, we always rushed out to buy the latest and greatest PC, because the newer models had faster cycles or more gigahertz. Moore's Law would double that performance every 24 months. Now that equates to about 40% annually. CPU performance is now moderated. That growth is now down to roughly 30% annual improvements. So technically speaking, Moore's Law as we know it was dead. But combined, if you look at the improvements in Apple's SoC since 2015, they've been on a pace that's higher than 118% annually. And it's even higher than that, because the actual figure for these three processor types we're not even counting the impact of DSPs and accelerator components of Apple system on a chip. It would push this even higher. Apple's A14 which is shown in the right hand side here is quite amazing. It's got a 64 bit architecture, it's got many, many cores. It's got a number of alternative processor types. But the important thing is what you can do with all this processing power. In an iPhone, the types of AI that we show here that continue to evolve, facial recognition, speech, natural language processing, rendering videos, helping the hearing impaired and eventually bringing augmented reality to the palm of your hand. It's quite incredible. So what does this mean for other parts of the IT stack? Well, we recently reported Satya Nadella's epic quote that "We've now reached peak centralization." So this graphic paints a picture that was quite telling. We just shared the processing powers exploding. The costs consequently are dropping like a rock. Apple's A14 cost the company approximately 50 bucks per chip. Arm at its v9 announcement said that it will have chips that can go into refrigerators. These chips are going to optimize energy usage and save 10% annually on your power consumption. They said, this chip will cost a buck, a dollar to shave 10% of your refrigerator electricity bill. It's just astounding. But look at where the expensive bottlenecks are, it's networks and it's storage. So what does this mean? Well, it means the processing is going to get pushed to the edge, i.e., wherever the data is born. Storage and networking are going to become increasingly distributed and decentralized. Now with custom silicon and all that processing power placed throughout the system, an AI is going to be embedded into software, into hardware and it's going to optimize a workloads for latency, performance, bandwidth, and security. And remember, most of that data, 99% is going to stay at the edge. And we love to use Tesla as an example. The vast majority of data that a Tesla car creates is never going to go back to the cloud. Most of it doesn't even get persisted. I think Tesla saves like five minutes of data. But some data will connect occasionally back to the cloud to train AI models and we're going to come back to that. But this picture says if you're a hardware company, you'd better start thinking about how to take advantage of that blue line that's exploding, Cisco. Cisco is already designing its own chips. But Dell, HPE, who kind of does maybe used to do a lot of its own custom silicon, but Pure Storage, NetApp, I mean, the list goes on and on and on either you're going to get start designing custom silicon or you're going to get disrupted in our view. AWS, Google and Microsoft are all doing it for a reason as is IBM and to Sarbjeet Johal said recently this is not your grandfather's semiconductor business. And if you're a software engineer, you're going to be writing applications that take advantage of all the data being collected and bringing to bear this processing power that we're talking about to create new capabilities like we've never seen it before. So let's get into that a little bit and dig into AI. You can think of AI as the superset. Just as an aside, interestingly in his book, "Seeing Digital", author David Moschella says, there's nothing artificial about this. He uses the term machine intelligence, instead of artificial intelligence and says that there's nothing artificial about machine intelligence just like there's nothing artificial about the strength of a tractor. It's a nuance, but it's kind of interesting, nonetheless, words matter. We hear a lot about machine learning and deep learning and think of them as subsets of AI. Machine learning applies algorithms and code to data to get "smarter", make better models, for example, that can lead to augmented intelligence and help humans make better decisions. These models improve as they get more data and are iterated over time. Now deep learning is a more advanced type of machine learning. It uses more complex math. But the point that we want to make here is that today much of the activity in AI is around building and training models. And this is mostly happening in the cloud. But we think AI inference will bring the most exciting innovations in the coming years. Inference is the deployment of that model that we were just talking about, taking real time data from sensors, processing that data locally and then applying that training that has been developed in the cloud and making micro adjustments in real time. So let's take an example. Again, we love Tesla examples. Think about an algorithm that optimizes the performance and safety of a car on a turn, the model take data on friction, road condition, angles of the tires, the tire wear, the tire pressure, all this data, and it keeps testing and iterating, testing and iterating, testing iterating that model until it's ready to be deployed. And then the intelligence, all this intelligence goes into an inference engine which is a chip that goes into a car and gets data from sensors and makes these micro adjustments in real time on steering and braking and the like. Now, as you said before, Tesla persist the data for very short time, because there's so much of it. It just can't push it back to the cloud. But it can now ever selectively store certain data if it needs to, and then send back that data to the cloud to further train them all. Let's say for instance, an animal runs into the road during slick conditions, Tesla wants to grab that data, because they notice that there's a lot of accidents in New England in certain months. And maybe Tesla takes that snapshot and sends it back to the cloud and combines it with other data and maybe other parts of the country or other regions of New England and it perfects that model further to improve safety. This is just one example of thousands and thousands that are going to further develop in the coming decade. I want to talk about how we see this evolving over time. Inference is where we think the value is. That's where the rubber meets the road, so to speak, based on the previous example. Now this conceptual chart shows the percent of spend over time on modeling versus inference. And you can see some of the applications that get attention today and how these applications will mature over time as inference becomes more and more mainstream, the opportunities for AI inference at the edge and in IOT are enormous. And we think that over time, 95% of that spending is going to go to inference where it's probably only 5% today. Now today's modeling workloads are pretty prevalent and things like fraud, adtech, weather, pricing, recommendation engines, and those kinds of things, and now those will keep getting better and better and better over time. Now in the middle here, we show the industries which are all going to be transformed by these trends. Now, one of the point that Moschella had made in his book, he kind of explains why historically vertically industries are pretty stovepiped, they have their own stack, sales and marketing and engineering and supply chains, et cetera, and experts within those industries tend to stay within those industries and they're largely insulated from disruption from other industries, maybe unless they were part of a supply chain. But today, you see all kinds of cross industry activity. Amazon entering grocery, entering media. Apple in finance and potentially getting into EV. Tesla, eyeing insurance. There are many, many, many examples of tech giants who are crossing traditional industry boundaries. And the reason is because of data. They have the data. And they're applying machine intelligence to that data and improving. Auto manufacturers, for example, over time they're going to have better data than insurance companies. DeFi, decentralized finance platforms going to use the blockchain and they're continuing to improve. Blockchain today is not great performance, it's very overhead intensive all that encryption. But as they take advantage of this new processing power and better software and AI, it could very well disrupt traditional payment systems. And again, so many examples here. But what I want to do now is dig into enterprise AI a bit. And just a quick reminder, we showed this last week in our Armv9 post. This is data from ETR. The vertical axis is net score. That's a measure of spending momentum. The horizontal axis is market share or pervasiveness in the dataset. The red line at 40% is like a subjective anchor that we use. Anything above 40% we think is really good. Machine learning and AI is the number one area of spending velocity and has been for awhile. RPA is right there. Very frankly, it's an adjacency to AI and you could even argue. So it's cloud where all the ML action is taking place today. But that will change, we think, as we just described, because data's going to get pushed to the edge. And this chart will show you some of the vendors in that space. These are the companies that CIOs and IT buyers associate with their AI and machine learning spend. So it's the same XY graph, spending velocity by market share on the horizontal axis. Microsoft, AWS, Google, of course, the big cloud guys they dominate AI and machine learning. Facebook's not on here. Facebook's got great AI as well, but it's not enterprise tech spending. These cloud companies they have the tooling, they have the data, they have the scale and as we said, lots of modeling is going on today, but this is going to increasingly be pushed into remote AI inference engines that will have massive processing capabilities collectively. So we're moving away from that peak centralization as Satya Nadella described. You see Databricks on here. They're seen as an AI leader. SparkCognition, they're off the charts, literally, in the upper left. They have extremely high net score albeit with a small sample. They apply machine learning to massive data sets. DataRobot does automated AI. They're super high in the y-axis. Dataiku, they help create machine learning based apps. C3.ai, you're hearing a lot more about them. Tom Siebel's involved in that company. It's an enterprise AI firm, hear a lot of ads now doing AI and responsible way really kind of enterprise AI that's sort of always been IBM. IBM Watson's calling card. There's SAP with Leonardo. Salesforce with Einstein. Again, IBM Watson is right there just at the 40% line. You see Oracle is there as well. They're embedding automated and tele or machine intelligence with their self-driving database they call it that sort of machine intelligence in the database. You see Adobe there. So a lot of typical enterprise company names. And the point is that these software companies they're all embedding AI into their offerings. So if you're an incumbent company and you're trying not to get disrupted, the good news is you can buy AI from these software companies. You don't have to build it. You don't have to be an expert at AI. The hard part is going to be how and where to apply AI. And the simplest answer there is follow the data. There's so much more to the story, but we just have to leave it there for now and I want to summarize. We have been pounding the table that the post x86 era is here. It's a function of volume. Arm volumes are a way for volumes are 10X those of x86. Pat Gelsinger understands this. That's why he made that big announcement. He's trying to transform the company. The importance of volume in terms of lowering the cost of semiconductors it can't be understated. And today, we've quantified something that we haven't really seen much of and really haven't seen before. And that's that the actual performance improvements that we're seeing in processing today are far outstripping anything we've seen before, forget Moore's Law being dead that's irrelevant. The original finding is being blown away this decade and who knows with quantum computing what the future holds. This is a fundamental enabler of AI applications. And this is most often the case the innovation is coming from the consumer use cases first. Apple continues to lead the way. And Apple's integrated hardware and software model we think increasingly is going to move into the enterprise mindset. Clearly the cloud vendors are moving in this direction, building their own custom silicon and doing really that deep integration. You see this with Oracle who kind of really a good example of the iPhone for the enterprise, if you will. It just makes sense that optimizing hardware and software together is going to gain momentum, because there's so much opportunity for customization in chips as we discussed last week with Arm's announcement, especially with the diversity of edge use cases. And it's the direction that Pat Gelsinger is taking Intel trying to provide more flexibility. One aside, Pat Gelsinger he may face massive challenges that we laid out a couple of posts ago with our Intel breaking analysis, but he is right on in our view that semiconductor demand is increasing. There's no end in sight. We don't think we're going to see these ebbs and flows as we've seen in the past that these boom and bust cycles for semiconductor. We just think that prices are coming down. The market's elastic and the market is absolutely exploding with huge demand for fab capacity. Now, if you're an enterprise, you should not stress about and trying to invent AI, rather you should put your focus on understanding what data gives you competitive advantage and how to apply machine intelligence and AI to win. You're going to be buying, not building AI and you're going to be applying it. Now data as John Furrier has said in the past is becoming the new development kit. He said that 10 years ago and he seems right. Finally, if you're an enterprise hardware player, you're going to be designing your own chips and writing more software to exploit AI. You'll be embedding custom silicon in AI throughout your product portfolio and storage and networking and you'll be increasingly bringing compute to the data. And that data will mostly stay where it's created. Again, systems and storage and networking stacks they're all being completely re-imagined. If you're a software developer, you now have processing capabilities in the palm of your hand that are incredible. And you're going to rewriting new applications to take advantage of this and use AI to change the world, literally. You'll have to figure out how to get access to the most relevant data. You have to figure out how to secure your platforms and innovate. And if you're a services company, your opportunity is to help customers that are trying not to get disrupted are many. You have the deep industry expertise and horizontal technology chops to help customers survive and thrive. Privacy? AI for good? Yeah well, that's a whole another topic. I think for now, we have to get a better understanding of how far AI can go before we determine how far it should go. Look, protecting our personal data and privacy should definitely be something that we're concerned about and we should protect. But generally, I'd rather not stifle innovation at this point. I'd be interested in what you think about that. Okay. That's it for today. Thanks to David Foyer, who helped me with this segment again and did a lot of the charts and the data behind this. He's done some great work there. Remember these episodes are all available as podcasts wherever you listen, just search breaking it analysis podcast and please subscribe to the series. We'd appreciate that. Check out ETR's website at ETR.plus. We also publish a full report with more detail every week on Wikibon.com and siliconangle.com, so check that out. You can get in touch with me. I'm dave.vellante@siliconangle.com. You can DM me on Twitter @dvellante or comment on our LinkedIn posts. I always appreciate that. This is Dave Vellante for theCUBE Insights powered by ETR. Stay safe, be well. And we'll see you next time. (bright music)

(upbeat music) >> Hello everyone. Welcome to theCube's presentation of the "AWS Startup Showcase." The topic this episode is AI and machine learning, top startups building foundational model infrastructure. This is season three, episode one of the ongoing series covering exciting startups from the AWS ecosystem. And this time we're talking about AI and machine learning. I'm your host, John Furrier. I'm excited I'm joined today by Robert Nishihara, who's the co-founder and CEO of a hot startup called Anyscale. He's here to talk about Ray, the open source project, Anyscale's infrastructure for foundation as well. Robert, thank you for joining us today. >> Yeah, thanks so much as well. >> I've been following your company since the founding pre pandemic and you guys really had a great vision scaled up and in a perfect position for this big wave that we all see with ChatGPT and OpenAI that's gone mainstream. Finally, AI has broken out through the ropes and now gone mainstream, so I think you guys are really well positioned. I'm looking forward to to talking with you today. But before we get into it, introduce the core mission for Anyscale. Why do you guys exist? What is the North Star for Anyscale? >> Yeah, like you mentioned, there's a tremendous amount of excitement about AI right now. You know, I think a lot of us believe that AI can transform just every different industry. So one of the things that was clear to us when we started this company was that the amount of compute needed to do AI was just exploding. Like to actually succeed with AI, companies like OpenAI or Google or you know, these companies getting a lot of value from AI, were not just running these machine learning models on their laptops or on a single machine. They were scaling these applications across hundreds or thousands or more machines and GPUs and other resources in the Cloud. And so to actually succeed with AI, and this has been one of the biggest trends in computing, maybe the biggest trend in computing in, you know, in recent history, the amount of compute has been exploding. And so to actually succeed with that AI, to actually build these scalable applications and scale the AI applications, there's a tremendous software engineering lift to build the infrastructure to actually run these scalable applications. And that's very hard to do. So one of the reasons many AI projects and initiatives fail is that, or don't make it to production, is the need for this scale, the infrastructure lift, to actually make it happen. So our goal here with Anyscale and Ray, is to make that easy, is to make scalable computing easy. So that as a developer or as a business, if you want to do AI, if you want to get value out of AI, all you need to know is how to program on your laptop. Like, all you need to know is how to program in Python. And if you can do that, then you're good to go. Then you can do what companies like OpenAI or Google do and get value out of machine learning. >> That programming example of how easy it is with Python reminds me of the early days of Cloud, when infrastructure as code was talked about was, it was just code the infrastructure programmable. That's super important. That's what AI people wanted, first program AI. That's the new trend. And I want to understand, if you don't mind explaining, the relationship that Anyscale has to these foundational models and particular the large language models, also called LLMs, was seen with like OpenAI and ChatGPT. Before you get into the relationship that you have with them, can you explain why the hype around foundational models? Why are people going crazy over foundational models? What is it and why is it so important? >> Yeah, so foundational models and foundation models are incredibly important because they enable businesses and developers to get value out of machine learning, to use machine learning off the shelf with these large models that have been trained on tons of data and that are useful out of the box. And then, of course, you know, as a business or as a developer, you can take those foundational models and repurpose them or fine tune them or adapt them to your specific use case and what you want to achieve. But it's much easier to do that than to train them from scratch. And I think there are three, for people to actually use foundation models, there are three main types of workloads or problems that need to be solved. One is training these foundation models in the first place, like actually creating them. The second is fine tuning them and adapting them to your use case. And the third is serving them and actually deploying them. Okay, so Ray and Anyscale are used for all of these three different workloads. Companies like OpenAI or Cohere that train large language models. Or open source versions like GPTJ are done on top of Ray. There are many startups and other businesses that fine tune, that, you know, don't want to train the large underlying foundation models, but that do want to fine tune them, do want to adapt them to their purposes, and build products around them and serve them, those are also using Ray and Anyscale for that fine tuning and that serving. And so the reason that Ray and Anyscale are important here is that, you know, building and using foundation models requires a huge scale. It requires a lot of data. It requires a lot of compute, GPUs, TPUs, other resources. And to actually take advantage of that and actually build these scalable applications, there's a lot of infrastructure that needs to happen under the hood. And so you can either use Ray and Anyscale to take care of that and manage the infrastructure and solve those infrastructure problems. Or you can build the infrastructure and manage the infrastructure yourself, which you can do, but it's going to slow your team down. It's going to, you know, many of the businesses we work with simply don't want to be in the business of managing infrastructure and building infrastructure. They want to focus on product development and move faster. >> I know you got a keynote presentation we're going to go to in a second, but I think you hit on something I think is the real tipping point, doing it yourself, hard to do. These are things where opportunities are and the Cloud did that with data centers. Turned a data center and made it an API. The heavy lifting went away and went to the Cloud so people could be more creative and build their product. In this case, build their creativity. Is that kind of what's the big deal? Is that kind of a big deal happening that you guys are taking the learnings and making that available so people don't have to do that? >> That's exactly right. So today, if you want to succeed with AI, if you want to use AI in your business, infrastructure work is on the critical path for doing that. To do AI, you have to build infrastructure. You have to figure out how to scale your applications. That's going to change. We're going to get to the point, and you know, with Ray and Anyscale, we're going to remove the infrastructure from the critical path so that as a developer or as a business, all you need to focus on is your application logic, what you want the the program to do, what you want your application to do, how you want the AI to actually interface with the rest of your product. Now the way that will happen is that Ray and Anyscale will still, the infrastructure work will still happen. It'll just be under the hood and taken care of by Ray in Anyscale. And so I think something like this is really necessary for AI to reach its potential, for AI to have the impact and the reach that we think it will, you have to make it easier to do. >> And just for clarification to point out, if you don't mind explaining the relationship of Ray and Anyscale real quick just before we get into the presentation. >> So Ray is an open source project. We created it. We were at Berkeley doing machine learning. We started Ray so that, in order to provide an easy, a simple open source tool for building and running scalable applications. And Anyscale is the managed version of Ray, basically we will run Ray for you in the Cloud, provide a lot of tools around the developer experience and managing the infrastructure and providing more performance and superior infrastructure. >> Awesome. I know you got a presentation on Ray and Anyscale and you guys are positioning as the infrastructure for foundational models. So I'll let you take it away and then when you're done presenting, we'll come back, I'll probably grill you with a few questions and then we'll close it out so take it away. >> Robert: Sounds great. So I'll say a little bit about how companies are using Ray and Anyscale for foundation models. The first thing I want to mention is just why we're doing this in the first place. And the underlying observation, the underlying trend here, and this is a plot from OpenAI, is that the amount of compute needed to do machine learning has been exploding. It's been growing at something like 35 times every 18 months. This is absolutely enormous. And other people have written papers measuring this trend and you get different numbers. But the point is, no matter how you slice and dice it, it' a astronomical rate. Now if you compare that to something we're all familiar with, like Moore's Law, which says that, you know, the processor performance doubles every roughly 18 months, you can see that there's just a tremendous gap between the needs, the compute needs of machine learning applications, and what you can do with a single chip, right. So even if Moore's Law were continuing strong and you know, doing what it used to be doing, even if that were the case, there would still be a tremendous gap between what you can do with the chip and what you need in order to do machine learning. And so given this graph, what we've seen, and what has been clear to us since we started this company, is that doing AI requires scaling. There's no way around it. It's not a nice to have, it's really a requirement. And so that led us to start Ray, which is the open source project that we started to make it easy to build these scalable Python applications and scalable machine learning applications. And since we started the project, it's been adopted by a tremendous number of companies. Companies like OpenAI, which use Ray to train their large models like ChatGPT, companies like Uber, which run all of their deep learning and classical machine learning on top of Ray, companies like Shopify or Spotify or Instacart or Lyft or Netflix, ByteDance, which use Ray for their machine learning infrastructure. Companies like Ant Group, which makes Alipay, you know, they use Ray across the board for fraud detection, for online learning, for detecting money laundering, you know, for graph processing, stream processing. Companies like Amazon, you know, run Ray at a tremendous scale and just petabytes of data every single day. And so the project has seen just enormous adoption since, over the past few years. And one of the most exciting use cases is really providing the infrastructure for building training, fine tuning, and serving foundation models. So I'll say a little bit about, you know, here are some examples of companies using Ray for foundation models. Cohere trains large language models. OpenAI also trains large language models. You can think about the workloads required there are things like supervised pre-training, also reinforcement learning from human feedback. So this is not only the regular supervised learning, but actually more complex reinforcement learning workloads that take human input about what response to a particular question, you know is better than a certain other response. And incorporating that into the learning. There's open source versions as well, like GPTJ also built on top of Ray as well as projects like Alpa coming out of UC Berkeley. So these are some of the examples of exciting projects in organizations, training and creating these large language models and serving them using Ray. Okay, so what actually is Ray? Well, there are two layers to Ray. At the lowest level, there's the core Ray system. This is essentially low level primitives for building scalable Python applications. Things like taking a Python function or a Python class and executing them in the cluster setting. So Ray core is extremely flexible and you can build arbitrary scalable applications on top of Ray. So on top of Ray, on top of the core system, what really gives Ray a lot of its power is this ecosystem of scalable libraries. So on top of the core system you have libraries, scalable libraries for ingesting and pre-processing data, for training your models, for fine tuning those models, for hyper parameter tuning, for doing batch processing and batch inference, for doing model serving and deployment, right. And a lot of the Ray users, the reason they like Ray is that they want to run multiple workloads. They want to train and serve their models, right. They want to load their data and feed that into training. And Ray provides common infrastructure for all of these different workloads. So this is a little overview of what Ray, the different components of Ray. So why do people choose to go with Ray? I think there are three main reasons. The first is the unified nature. The fact that it is common infrastructure for scaling arbitrary workloads, from data ingest to pre-processing to training to inference and serving, right. This also includes the fact that it's future proof. AI is incredibly fast moving. And so many people, many companies that have built their own machine learning infrastructure and standardized on particular workflows for doing machine learning have found that their workflows are too rigid to enable new capabilities. If they want to do reinforcement learning, if they want to use graph neural networks, they don't have a way of doing that with their standard tooling. And so Ray, being future proof and being flexible and general gives them that ability. Another reason people choose Ray in Anyscale is the scalability. This is really our bread and butter. This is the reason, the whole point of Ray, you know, making it easy to go from your laptop to running on thousands of GPUs, making it easy to scale your development workloads and run them in production, making it easy to scale, you know, training to scale data ingest, pre-processing and so on. So scalability and performance, you know, are critical for doing machine learning and that is something that Ray provides out of the box. And lastly, Ray is an open ecosystem. You can run it anywhere. You can run it on any Cloud provider. Google, you know, Google Cloud, AWS, Asure. You can run it on your Kubernetes cluster. You can run it on your laptop. It's extremely portable. And not only that, it's framework agnostic. You can use Ray to scale arbitrary Python workloads. You can use it to scale and it integrates with libraries like TensorFlow or PyTorch or JAX or XG Boost or Hugging Face or PyTorch Lightning, right, or Scikit-learn or just your own arbitrary Python code. It's open source. And in addition to integrating with the rest of the machine learning ecosystem and these machine learning frameworks, you can use Ray along with all of the other tooling in the machine learning ecosystem. That's things like weights and biases or ML flow, right. Or you know, different data platforms like Databricks, you know, Delta Lake or Snowflake or tools for model monitoring for feature stores, all of these integrate with Ray. And that's, you know, Ray provides that kind of flexibility so that you can integrate it into the rest of your workflow. And then Anyscale is the scalable compute platform that's built on top, you know, that provides Ray. So Anyscale is a managed Ray service that runs in the Cloud. And what Anyscale does is it offers the best way to run Ray. And if you think about what you get with Anyscale, there are fundamentally two things. One is about moving faster, accelerating the time to market. And you get that by having the managed service so that as a developer you don't have to worry about managing infrastructure, you don't have to worry about configuring infrastructure. You also, it provides, you know, optimized developer workflows. Things like easily moving from development to production, things like having the observability tooling, the debug ability to actually easily diagnose what's going wrong in a distributed application. So things like the dashboards and the other other kinds of tooling for collaboration, for monitoring and so on. And then on top of that, so that's the first bucket, developer productivity, moving faster, faster experimentation and iteration. The second reason that people choose Anyscale is superior infrastructure. So this is things like, you know, cost deficiency, being able to easily take advantage of spot instances, being able to get higher GPU utilization, things like faster cluster startup times and auto scaling. Things like just overall better performance and faster scheduling. And so these are the kinds of things that Anyscale provides on top of Ray. It's the managed infrastructure. It's fast, it's like the developer productivity and velocity as well as performance. So this is what I wanted to share about Ray in Anyscale. >> John: Awesome. >> Provide that context. But John, I'm curious what you think. >> I love it. I love the, so first of all, it's a platform because that's the platform architecture right there. So just to clarify, this is an Anyscale platform, not- >> That's right. >> Tools. So you got tools in the platform. Okay, that's key. Love that managed service. Just curious, you mentioned Python multiple times, is that because of PyTorch and TensorFlow or Python's the most friendly with machine learning or it's because it's very common amongst all developers? >> That's a great question. Python is the language that people are using to do machine learning. So it's the natural starting point. Now, of course, Ray is actually designed in a language agnostic way and there are companies out there that use Ray to build scalable Java applications. But for the most part right now we're focused on Python and being the best way to build these scalable Python and machine learning applications. But, of course, down the road there always is that potential. >> So if you're slinging Python code out there and you're watching that, you're watching this video, get on Anyscale bus quickly. Also, I just, while you were giving the presentation, I couldn't help, since you mentioned OpenAI, which by the way, congratulations 'cause they've had great scale, I've noticed in their rapid growth 'cause they were the fastest company to the number of users than anyone in the history of the computer industry, so major successor, OpenAI and ChatGPT, huge fan. I'm not a skeptic at all. I think it's just the beginning, so congratulations. But I actually typed into ChatGPT, what are the top three benefits of Anyscale and came up with scalability, flexibility, and ease of use. Obviously, scalability is what you guys are called. >> That's pretty good. >> So that's what they came up with. So they nailed it. Did you have an inside prompt training, buy it there? Only kidding. (Robert laughs) >> Yeah, we hard coded that one. >> But that's the kind of thing that came up really, really quickly if I asked it to write a sales document, it probably will, but this is the future interface. This is why people are getting excited about the foundational models and the large language models because it's allowing the interface with the user, the consumer, to be more human, more natural. And this is clearly will be in every application in the future. >> Absolutely. This is how people are going to interface with software, how they're going to interface with products in the future. It's not just something, you know, not just a chat bot that you talk to. This is going to be how you get things done, right. How you use your web browser or how you use, you know, how you use Photoshop or how you use other products. Like you're not going to spend hours learning all the APIs and how to use them. You're going to talk to it and tell it what you want it to do. And of course, you know, if it doesn't understand it, it's going to ask clarifying questions. You're going to have a conversation and then it'll figure it out. >> This is going to be one of those things, we're going to look back at this time Robert and saying, "Yeah, from that company, that was the beginning of that wave." And just like AWS and Cloud Computing, the folks who got in early really were in position when say the pandemic came. So getting in early is a good thing and that's what everyone's talking about is getting in early and playing around, maybe replatforming or even picking one or few apps to refactor with some staff and managed services. So people are definitely jumping in. So I have to ask you the ROI cost question. You mentioned some of those, Moore's Law versus what's going on in the industry. When you look at that kind of scale, the first thing that jumps out at people is, "Okay, I love it. Let's go play around." But what's it going to cost me? Am I going to be tied to certain GPUs? What's the landscape look like from an operational standpoint, from the customer? Are they locked in and the benefit was flexibility, are you flexible to handle any Cloud? What is the customers, what are they looking at? Basically, that's my question. What's the customer looking at? >> Cost is super important here and many of the companies, I mean, companies are spending a huge amount on their Cloud computing, on AWS, and on doing AI, right. And I think a lot of the advantage of Anyscale, what we can provide here is not only better performance, but cost efficiency. Because if we can run something faster and more efficiently, it can also use less resources and you can lower your Cloud spending, right. We've seen companies go from, you know, 20% GPU utilization with their current setup and the current tools they're using to running on Anyscale and getting more like 95, you know, 100% GPU utilization. That's something like a five x improvement right there. So depending on the kind of application you're running, you know, it's a significant cost savings. We've seen companies that have, you know, processing petabytes of data every single day with Ray going from, you know, getting order of magnitude cost savings by switching from what they were previously doing to running their application on Ray. And when you have applications that are spending, you know, potentially $100 million a year and getting a 10 X cost savings is just absolutely enormous. So these are some of the kinds of- >> Data infrastructure is super important. Again, if the customer, if you're a prospect to this and thinking about going in here, just like the Cloud, you got infrastructure, you got the platform, you got SaaS, same kind of thing's going to go on in AI. So I want to get into that, you know, ROI discussion and some of the impact with your customers that are leveraging the platform. But first I hear you got a demo. >> Robert: Yeah, so let me show you, let me give you a quick run through here. So what I have open here is the Anyscale UI. I've started a little Anyscale Workspace. So Workspaces are the Anyscale concept for interactive developments, right. So here, imagine I'm just, you want to have a familiar experience like you're developing on your laptop. And here I have a terminal. It's not on my laptop. It's actually in the cloud running on Anyscale. And I'm just going to kick this off. This is going to train a large language model, so OPT. And it's doing this on 32 GPUs. We've got a cluster here with a bunch of CPU cores, bunch of memory. And as that's running, and by the way, if I wanted to run this on instead of 32 GPUs, 64, 128, this is just a one line change when I launch the Workspace. And what I can do is I can pull up VS code, right. Remember this is the interactive development experience. I can look at the actual code. Here it's using Ray train to train the torch model. We've got the training loop and we're saying that each worker gets access to one GPU and four CPU cores. And, of course, as I make the model larger, this is using deep speed, as I make the model larger, I could increase the number of GPUs that each worker gets access to, right. And how that is distributed across the cluster. And if I wanted to run on CPUs instead of GPUs or a different, you know, accelerator type, again, this is just a one line change. And here we're using Ray train to train the models, just taking my vanilla PyTorch model using Hugging Face and then scaling that across a bunch of GPUs. And, of course, if I want to look at the dashboard, I can go to the Ray dashboard. There are a bunch of different visualizations I can look at. I can look at the GPU utilization. I can look at, you know, the CPU utilization here where I think we're currently loading the model and running that actual application to start the training. And some of the things that are really convenient here about Anyscale, both I can get that interactive development experience with VS code. You know, I can look at the dashboards. I can monitor what's going on. It feels, I have a terminal, it feels like my laptop, but it's actually running on a large cluster. And I can, with however many GPUs or other resources that I want. And so it's really trying to combine the best of having the familiar experience of programming on your laptop, but with the benefits, you know, being able to take advantage of all the resources in the Cloud to scale. And it's like when, you know, you're talking about cost efficiency. One of the biggest reasons that people waste money, one of the silly reasons for wasting money is just forgetting to turn off your GPUs. And what you can do here is, of course, things will auto terminate if they're idle. But imagine you go to sleep, I have this big cluster. You can turn it off, shut off the cluster, come back tomorrow, restart the Workspace, and you know, your big cluster is back up and all of your code changes are still there. All of your local file edits. It's like you just closed your laptop and came back and opened it up again. And so this is the kind of experience we want to provide for our users. So that's what I wanted to share with you. >> Well, I think that whole, couple of things, lines of code change, single line of code change, that's game changing. And then the cost thing, I mean human error is a big deal. People pass out at their computer. They've been coding all night or they just forget about it. I mean, and then it's just like leaving the lights on or your water running in your house. It's just, at the scale that it is, the numbers will add up. That's a huge deal. So I think, you know, compute back in the old days, there's no compute. Okay, it's just compute sitting there idle. But you know, data cranking the models is doing, that's a big point. >> Another thing I want to add there about cost efficiency is that we make it really easy to use, if you're running on Anyscale, to use spot instances and these preemptable instances that can just be significantly cheaper than the on-demand instances. And so when we see our customers go from what they're doing before to using Anyscale and they go from not using these spot instances 'cause they don't have the infrastructure around it, the fault tolerance to handle the preemption and things like that, to being able to just check a box and use spot instances and save a bunch of money. >> You know, this was my whole, my feature article at Reinvent last year when I met with Adam Selipsky, this next gen Cloud is here. I mean, it's not auto scale, it's infrastructure scale. It's agility. It's flexibility. I think this is where the world needs to go. Almost what DevOps did for Cloud and what you were showing me that demo had this whole SRE vibe. And remember Google had site reliability engines to manage all those servers. This is kind of like an SRE vibe for data at scale. I mean, a similar kind of order of magnitude. I mean, I might be a little bit off base there, but how would you explain it? >> It's a nice analogy. I mean, what we are trying to do here is get to the point where developers don't think about infrastructure. Where developers only think about their application logic. And where businesses can do AI, can succeed with AI, and build these scalable applications, but they don't have to build, you know, an infrastructure team. They don't have to develop that expertise. They don't have to invest years in building their internal machine learning infrastructure. They can just focus on the Python code, on their application logic, and run the stuff out of the box. >> Awesome. Well, I appreciate the time. Before we wrap up here, give a plug for the company. I know you got a couple websites. Again, go, Ray's got its own website. You got Anyscale. You got an event coming up. Give a plug for the company looking to hire. Put a plug in for the company. >> Yeah, absolutely. Thank you. So first of all, you know, we think AI is really going to transform every industry and the opportunity is there, right. We can be the infrastructure that enables all of that to happen, that makes it easy for companies to succeed with AI, and get value out of AI. Now we have, if you're interested in learning more about Ray, Ray has been emerging as the standard way to build scalable applications. Our adoption has been exploding. I mentioned companies like OpenAI using Ray to train their models. But really across the board companies like Netflix and Cruise and Instacart and Lyft and Uber, you know, just among tech companies. It's across every industry. You know, gaming companies, agriculture, you know, farming, robotics, drug discovery, you know, FinTech, we see it across the board. And all of these companies can get value out of AI, can really use AI to improve their businesses. So if you're interested in learning more about Ray and Anyscale, we have our Ray Summit coming up in September. This is going to highlight a lot of the most impressive use cases and stories across the industry. And if your business, if you want to use LLMs, you want to train these LLMs, these large language models, you want to fine tune them with your data, you want to deploy them, serve them, and build applications and products around them, give us a call, talk to us. You know, we can really take the infrastructure piece, you know, off the critical path and make that easy for you. So that's what I would say. And, you know, like you mentioned, we're hiring across the board, you know, engineering, product, go-to-market, and it's an exciting time. >> Robert Nishihara, co-founder and CEO of Anyscale, congratulations on a great company you've built and continuing to iterate on and you got growth ahead of you, you got a tailwind. I mean, the AI wave is here. I think OpenAI and ChatGPT, a customer of yours, have really opened up the mainstream visibility into this new generation of applications, user interface, roll of data, large scale, how to make that programmable so we're going to need that infrastructure. So thanks for coming on this season three, episode one of the ongoing series of the hot startups. In this case, this episode is the top startups building foundational model infrastructure for AI and ML. I'm John Furrier, your host. Thanks for watching. (upbeat music)

(upbeat music) >> For decades the technology industry had marched the cadence of Moore's law. It was a familiar pattern. System OEMs would design in the next generation of Intel microprocessors, every couple of years or so maybe bump up the memory ranges periodically and the supporting hardware would kind of go along for the ride, upgrading its performance and bandwidth. System designers then they might beef up the cache, maybe throwing some more spinning disc spindles at the equation to create a balanced environment. And this was pretty predictable and consistent in the pattern and was reasonably straightforward compared to is challenges. This has all changed. The confluence of cloud, distributed global networks, the diversity of applications, AI, machine learning and the massive growth of data outside of the data center requires new architectures to keep up. As we've reported the traditional Moore's Law curve is flattening. And along with that we've seen new packages with alternative processors like GPUs, NPUs, accelerators and the like and the rising importance of supporting hardware to offload tasks like storage and security. And it's created a massive challenge to connect all these components together, the storage, the memories and all of the enabling hardware and do so securely at very low latency at scale and of course, cost effectively. This is the topic of today's segment. The shift from a world that is CPU centric to one where the connectivity of the various hardware components is where much of the innovation is occurring. And to talk about that, there is no company who knows more about out this topic than Broadcom. And with us today is Jas Tremblay, who is general manager, data center solutions group at Broadcom. Jas, welcome to theCUBE. >> Hey Dave, thanks for having me, really appreciate it. >> Yeah, you bet. Now Broadcom is a company that a lot of people might not know about. I mean, but the vast majority of the internet traffic flows through Broadcom products. (chuckles) Like pretty much all of it. It's a company with trailing 12 month revenues of nearly 29 billion and a 240 billion market cap. Jas, what else should people know about Broadcom? >> Well, Dave, 99% of the internet traffic goes through Broadcom silicon or devices. And I think what people are not often aware of is how breadth it is. It starts with the devices, phones and tablets that use our Wi-Fi technology or RF filters. And then those connect to access points either at home, at work or public access points using our Wi-Fi technology. And if you're working from home, you're using a residential or broadband gateway and that uses Broadcom technology also. From there you go to access networks, core networks and eventually you'll work your way into the data center, all connected by Broadcom. So really we're at the heart of enabling this connectivity ecosystem and we're at the core of it, we're a technology company. We invest about 5 billion a year in R&D. And as you were saying our last year we achieved 27.5 billion of revenue. And our mission is really to connect the ecosystem to enable what you said, this transformation around the data-centric world. >> So talk about your scope of responsibility. What's your role generally and specifically with storage? >> So I've been with the company for 16 years and I head up the data center solutions group which includes three product franchises PCA fabric, storage connectivity and Broadcom ethernet nics. So my charter, my team's charter is really server connectivity inside the data center. >> And what specifically is Broadcom doing in storage, Jas? >> So it's been quite a journey. Over the past eight years we've made a series of acquisition and built up a pretty impressive storage portfolio. This first started with LSI and that's where I came from. And the team here came from LSI that had two product franchises around storage. The first one was server connectivity, HBA raid, expanders for SSDs and HDDs. The second product group was actually chips that go inside the hard drives. So SOCs and pre amps. So that was an acquisition that we made and actually that's how I came into the Broadcom group through LSI. The next acquisition we made was PLX, the industry's leader in PCIe fabrics. They'd been doing PCIe switches for about 15 years. We acquired the company and really saw an acceleration in the requirements for NVMe attached and AI ML fabrics, very specialized, low latency fabrics. After that, we acquired a large system and software company, Brocade, and Dave if you recall, Brocade they're the market leader in fiber channel switching, this is where if you're financial or government institution you want to build a mission critical, ultra secure really best in class storage network. Following Brocade acquisition we acquired Emulex that is now the number one provider of fiber channel adapters inside servers. And the last acquisition for this puzzle was actually Broadcom where Avago acquired Broadcom and took on the Broadcom name. And there we acquired ethernet switching capabilities and ethernet adapters that go into storage servers or external storage systems. So with all this it's been quite the journey to build up this portfolio. We're number one in each of these storage product categories. And we now have four divisions that are focused on storage connectivity. >> That's quite remarkable when you think about it. I mean, I know all these companies that you were talking about and they were very quality companies but they were kind of bespoke in the fact that you had the vision to kind of connect the dots and now take responsibility for that integration. We're going to talk about what that means in terms of competitive advantage, but I wonder if we could zoom out and maybe you could talk about the key storage challenges and elaborate a little bit on why connectivity is now so important. Like what are the trends that are driving that shift that we talked about earlier from a CPU centric world to one that's connectivity centric? >> I think at Broadcom, we recognize the importance of storage and storage connectivity. And if you look at data centers whether it be private, public cloud or hybrid data centers, they're getting inundated with data. If you look at the digital universe it's growing at about 23% a day. So over a course of four to five years you're doubling the amount of new information and that poses really two key challenges for the infrastructure. The first one is you have to take all this data and for a good chunk of it, you have to store it, be able to access it and protect it. The second challenge is you actually have to go and analyze and process this data and doing this at scale that's the key challenge and what we're seeing these data centers getting a tsunami of data. And historically they've been CPU centric architectures. And what that means is the CPU's at the heart of the data center. And a lot of the workloads are processed by software running on the CPU. We believe that we're currently transforming the architecture from CPU centric to connectivity centric. And what we mean by connectivity centric is you architect your data center thinking about the connectivity first. And the goal of the connectivity is to use all the components inside the data center, the memory, the spinning media, the flash storage, the networking, the specialized accelerators, the FPGA all these elements and use them for what they're best at to process all this data. And the goal Dave is really to drive down power and deliver the performance so that we can achieve all the innovation we want inside the data centers. So it's really a shift from CPU centric to bringing in more specialized components and architecting the connectivity inside the data center to help. We think that's a really important part. >> So you have this need for connectivity at scale, you mentioned, and you're dealing with massive, massive amounts of data. I mean, we're going to look back to the last decade and say, oh, you've seen nothing compared to when we get to 2030, but at the same time you have to control costs. So what are the technical challenges to achieving that vision? >> So it's really challenging. It's not that complex to build up faster, bigger solution, if you have no cost or power budget. And really the key challenges that our team is facing working with customers is first, I'd say it's architectural challenges. So we would all like to have one fabric that aim to connect all the devices and bring us all the characteristics that we need. But the reality is, we can't do that. So you need distinct fabrics inside the data center and you need them to work together. You'll need an ethernet backbone. In some cases, you'll need a fiber channel network. In some cases, you'll need a small fabric for thousands or hundreds of thousands of HDDs. You will need PCIe fabrics for AI ML servers. And one of the key architectural challenges is which fabric do you use when and how do you develop these fabrics to meet their purpose built needs. That's one thing. The second architectural challenge, Dave is what I challenge my team with is example, how do I double bandwidth while reducing net power, double bandwidth, reducing net power? How do I take a storage controller and increase the IOPS by 10X and will allocate only 50% more power budget? So that equation requires tremendous innovation. And that's really what we focus on and power is becoming more and more important in that equation. So you've got decisions from an architecture perspective as to which fabric to use. You've got this architectural challenge around we need to innovate and do things smarter, better, to drive down power while delivering more performance. Then if you take those things together the problem statement becomes more complex. So you've had these silicon devices with complex firmware on them that need to inter-operate with multiple devices. They're getting more and more complex. So there's execution challenges and what we need to do. And what we're we're investing to do is shift left quality. So to do these complex devices that they come out time to market with high quality. And one of the key things Dave that we've invested in is emulation of the environment before you tape out your silicon. So effectively taking the application software, running it on an emulation environment, making sure that works, running your tests before you tape out and that ensures quality silicon. So it's challenging, but the team loves challenges. And that's kind of what we're facing, on one hand architectural challenges, on the other hand a new level of execution challenges. >> So you're compressing the time to final tape out versus maybe traditional techniques. And then, you mentioned architecture, am I right Jas that you're essentially from an architectural standpoint trying to minimize the... 'cause your latency's so important you're trying to minimize the amount of data that you have to move around and actually bringing compute to the data. Is that the right way to think about it? >> Well, I think that there's multiple parts of the problem. One of them is you need to do more data transactions, example data protection with rate algorithms. We need to do millions of transactions per second. And the only way to achieve this with the minimal power impact is to hardware accelerate these. That's one piece of investment. The other investment is, you're absolutely right, Dave. So it's shuffling the data around the data center. So in the data center in some cases you need to have multiple pieces of the puzzle, multiple ingredients processing the same data at the same time and you need advanced methodologies to share the data and avoid moving it all over the data center. So that's another big piece of investment that we're focused on. >> So let's stay on that because I see this as disruptive. You talk about spending $5 billion a year in R&D and talk a little bit more about the disruptive technologies or the supportive technologies that you're introducing specifically to support this vision. >> So let's break it down in a couple big industry problems that our team is focused on. So the first one is I'll take an enterprise workload database. If you want the fastest running database you want to utilize local storage and NVMe based drives and you need to protect that data. And raid is the mechanism of choice to protect your data in local environments. And there what we need to do is really just do the transactions a lot faster. Historically the storage has been a bit of a bottleneck in these types of applications. So example our newest generation product. We're doubling the bandwidth, increasing IOPS by four X, but more importantly we're accelerating raid rebuilds by 50X. And that's an important Dave, if you are using a database in some cases, you limit the size of that database based on how fast you can do those rebuilds. So this 50X acceleration in rebuilds is something we're getting a lot of good feedback on for customers. The last metric we're really focused on is write latency. So how fast can the CPU send the write to the storage connectivity subsystem and committed to drives? And we're improving that by 60X generation over generation. So we're talking fully loaded latency, 10 microseconds. So from an enterprise workload it's about data protection, much, much faster using NVMe drives. That's one big problem. The other one is if you look at Dave YouTube, Facebook, TikTok the amount of user generated content specifically video content that they're producing on an hour by hour basis is mind boggling. And the hyperscale customers are really counting on us to help them scale the connectivity of hundreds of thousands of hard drive to store and access all that data in a very reliable way. So there we're leading the industry in the transition to 24 gig SaaS and multi actuator drives. Third big problem is around AI ML servers. So these are some of the highest performance servers, that they basically need super low latency connectivity between GPGPUs, networking, NVMe drives, CPUs and orchestrate that all together. And the fabric of choice for that is PCIe fabric. So here, we're talking about 115 nanosecond latency in a PCIe fabric, fully nonblocking, very reliable. And here we're helping the industry transition from PCA gen four to PCIe gen five. And the last piece is okay, I've got a AI ML server, I have a storage system with hard drives or a storage server in the enterprise space. All these devices, systems need to be connected to the ethernet backbone. And my team is heavily investing in ethernet mix transitioning to 100 gig, 200 gig, 400 gig and putting capabilities optimized for storage workloads. So those are kind of the four big things that we're focused on at the industry level, from a connectivity perspective, Dave. >> And that makes a lot of sense and really resonates particularly as we have that shift from a CPU centric to a connectivity centric. And the other thing you said, I mean, you're talking about 50X rate rebuild times, a couple of things you know in storage is if you ask the question, what happens when something goes wrong? 'Cause it's all about recovery, you can't lose data. And the other thing you mentioned is write latency, which has always been the problem. Okay, reads, I can read out cache but ultimately you've got to get it to where it's persisted. So some real technical challenges there that you guys are dealing with. >> Absolutely, Dave. And these are the type of problems that gets the engineers excited. Give them really tough technical problems to go solve. >> I wonder if we could take a couple of examples or an example of scaling with a large customer, for instance obviously hyperscalers or take a company like Dell. I mean they're big company, big customer. Take us through that. >> So we use the word scale a lot at Broadcom. We work with some of the industry leaders and data centers and OEMs and scale means different things to them. So example, if I'm working with a hyperscaler that is getting inundated with data and they need half a million storage controllers to store all that data, well their scale problem is, can you deliver? And Dave, you know how much of a hot topic that is these days. So they need a partner that can scale from a delivery perspective. But if I take a company like example Dell that's very focused on storage, from storage servers, their acquisition of EMC. They have a very broad portfolio of data center storage offerings and scale to them from a connected by Broadcom perspective means that you need to have the investment scale to meet their end to end requirements. All the way from a low end storage connectivity solution for booting a server all the way up to a very high end all flash array or high density HDD system. So they want a company a partner that can invest and has a scale to invest to meet their end to end requirements. Second thing is their different products are unique and have different requirements and you need to adapt your collaboration model. So example, some products within Dell portfolio might say, I just want a storage adaptor, plug it in, the operating system will automatically recognize it. I need this turnkey. I want to do minimal investment, is not an area of high differentiation for me. At the other end of the spectrum they may have applications where they want deep integration with their management and our silicon tools so that they can deliver the highest quality, highest performance to their customers. So they need a partner that can scale from an R&D investment perspective from silicon software and hardware perspective but they also need a company that can scale from support and business model perspective and give them the flexibility that their end customers need. So Dell is a great company to work with. We have a long lasting relationship with them and the relationship is very deep in some areas, example server storage, and is also quite broad. They are adopters of the vast majority of our storage connectivity products. >> Well, and I imagine it was. Well I want to talk about the uniqueness of Broadcom again, I'm in awe of the fact that somebody had the vision, you guys, your team obviously your CEO was one of the visionaries of the industry, had the sense to look out and say, okay, we can put these pieces together. So I would imagine a company like Dell, they're able to consolidate their vendor their supplier base and push you for integration and innovation. How unique is the Broadcom model? What's compelling to your customer about that model? >> So I think what's unique from a storage perspective is the breadth of the portfolio and also the scale at which we can invest. So if you look at some of the things we talked about from a scale perspective how data centers throughout the world are getting inundated with data, Dave, they need help. And we need to equip them with cutting edge technology to increase performance, drive down power, improve reliability. So they need partners that in each of the product categories that you partner with them on, we can invest with scale. So that's, I think one of the first things. The second thing is, if you look at this connectivity centric data center you need multiple types of fabric. And whether it be cloud customers or large OEMs they are organizing themselves to be able to look at things holistically. They're no longer product company, they're very data center architecture companies. And so it's good for them to have a partner that can look across product groups across divisions says, okay this is the innovation we need to bring to market. These are the problems we need to go solve and they really appreciate that. And I think the last thing is a flexible business model. Within example, my division, we offer different business models, different engagement and collaboration models with technology. But there's another division that if you want to innovate at the silicon level and build custom silicon for you like many of the hyperscalers or other companies are doing that division is just focus on that. So I feel like Broadcom is unique from a storage perspective it's ability to innovate, breadth of portfolio and the flexibility in the collaboration model to help our customers solve their customers problems. >> So you're saying you can deal with merchant products slash open products or you can do high customization. Where does software differentiation fit into this model? >> So it's actually one of the most important elements. I think a lot of our customers take it for granted that will take care of the silicon will anticipate the requirements and deliver the performance that they need, but from a software, firmware, driver, utilities that is where a lot of differentiation lies. Some cases we'll offer an SDK model where customers can build their entire applications on top of that. In some cases they want to complete turnkey solution where you take technology, integrate it into server and the operating system recognizes it and you have outer box drivers from Broadcom. So we need to offer them that flexibility because their needs are quite broad there. >> So last question, what's the future of the business look like to Jas Tremblay? Give us your point of view on that. >> Well, it's fun. I got to tell you, Dave, we're having a great time. I've got a great team, they're the world's experts on storage connectivity and working with them is a pleasure. And we've got a rich, great set of customers that are giving us cool problems to go solve and we're excited about it. So I think this is really, with the acceleration of all this digital transformation that we're seeing, we're excited, we're having fun. And I think there's a lot of problems to be solved. And we also have a responsibility. I think the ecosystem and the industry is counting on our team to deliver the innovation from a storage connectivity perspective. And I'll tell you, Dave, we're having fun. It's great but we take that responsibility pretty seriously. >> Jas, great stuff. I really appreciate you laying all that out. Very important role you guys are playing. You have a really unique perspective. Thank you. >> Thank you, Dave. >> And thank you for watching. This is Dave Vellante for theCUBE and we'll see you next time.

>> From theCUBE Studios in Palo Alto in Boston, bringing you data-driven insights from theCUBE and ETR, this is "Breaking Analysis" with Dave Vellante. >> If it weren't for Pat Gelsinger, Intel's future would be a disaster. Even with his clear vision, fantastic leadership, deep technical and business acumen, and amazing positivity, the company's future is in serious jeopardy. It's the same story we've been telling for years. Volume is king in the semiconductor industry, and Intel no longer is the volume leader. Despite Intel's efforts to change that dynamic With several recent moves, including making another go at its Foundry business, the company is years away from reversing its lagging position relative to today's leading foundries and design shops. Intel's best chance to survive as a leader in our view, will come from a combination of a massive market, continued supply constraints, government money, and luck, perhaps in the form of a deal with apple in the midterm. Hello, and welcome to this week's "Wikibon CUBE Insights, Powered by ETR." In this "Breaking Analysis," we'll update you on our latest assessment of Intel's competitive position and unpack nuggets from the company's February investor conference. Let's go back in history a bit and review what we said in the early 2010s. If you've followed this program, you know that our David Floyer sounded the alarm for Intel as far back as 2012, the year after PC volumes peaked. Yes, they've ticked up a bit in the past couple of years but they pale in comparison to the volumes that the ARM ecosystem is producing. The world has changed from people entering data into machines, and now it's machines that are driving all the data. Data volumes in Web 1.0 were largely driven by keystrokes and clicks. Web 3.0 is going to be driven by machines entering data into sensors, cameras. Other edge devices are going to drive enormous data volumes and processing power to boot. Every windmill, every factory device, every consumer device, every car, will require processing at the edge to run AI, facial recognition, inference, and data intensive workloads. And the volume of this space compared to PCs and even the iPhone itself is about to be dwarfed with an explosion of devices. Intel is not well positioned for this new world in our view. Intel has to catch up on the process, Intel has to catch up on architecture, Intel has to play catch up on security, Intel has to play catch up on volume. The ARM ecosystem has cumulatively shipped 200 billion chips to date, and is shipping 10x Intel's wafer volume. Intel has to have an architecture that accommodates much more diversity. And while it's working on that, it's years behind. All that said, Pat Gelsinger is doing everything he can and more to close the gap. Here's a partial list of the moves that Pat is making. A year ago, he announced IDM 2.0, a new integrated device manufacturing strategy that opened up its world to partners for manufacturing and other innovation. Intel has restructured, reorganized, and many executives have boomeranged back in, many previous Intel execs. They understand the business and have a deep passion to help the company regain its prominence. As part of the IDM 2.0 announcement, Intel created, recreated if you will, a Foundry division and recently acquired Tower Semiconductor an Israeli firm, that is going to help it in that mission. It's opening up partnerships with alternative processor manufacturers and designers. And the company has announced major investments in CAPEX to build out Foundry capacity. Intel is going to spin out Mobileye, a company it had acquired for 15 billion in 2017. Or does it try and get a $50 billion valuation? Mobileye is about $1.4 billion in revenue, and is likely going to be worth more around 25 to 30 billion, we'll see. But Intel is going to maybe get $10 billion in cash from that, that spin out that IPO and it can use that to fund more FABS and more equipment. Intel is leveraging its 19,000 software engineers to move up the stack and sell more subscriptions and high margin software. He got to sell what he got. And finally Pat is playing politics beautifully. Announcing for example, FAB investments in Ohio, which he dubbed Silicon Heartland. Brilliant! Again, there's no doubt that Pat is moving fast and doing the right things. Here's Pat at his investor event in a T-shirt that says, "torrid, bringing back the torrid pace and discipline that Intel is used to." And on the right is Pat at the State of the Union address, looking sharp in shirt and tie and suit. And he has said, "a bet on Intel is a hedge against geopolitical instability in the world." That's just so good. To that statement, he showed this chart at his investor meeting. Basically it shows that whereas semiconductor manufacturing capacity has gone from 80% of the world's volume to 20%, he wants to get it back to 50% by 2030, and reset supply chains in a market that has become important as oil. Again, just brilliant positioning and pushing all the right hot buttons. And here's a slide underscoring that commitment, showing manufacturing facilities around the world with new capacity coming online in the next few years in Ohio and the EU. Mentioning the CHIPS Act in his presentation in The US and Europe as part of a public private partnership, no doubt, he's going to need all the help he can get. Now, we couldn't resist the chart on the left here shows wafer starts and transistor capacity growth. For Intel, overtime speaks to its volume aspirations. But we couldn't help notice that the shape of the curve is somewhat misleading because it shows a two-year (mumbles) and then widens the aperture to three years to make the curve look steeper. Fun with numbers. Okay, maybe a little nitpick, but these are some of the telling nuggets we pulled from the investor day, and they're important. Another nitpick is in our view, wafers would be a better measure of volume than transistors. It's like a company saying we shipped 20% more exabytes or MIPS this year than last year. Of course you did, and your revenue shrank. Anyway, Pat went through a detailed analysis of the various Intel businesses and promised mid to high double digit growth by 2026, half of which will come from Intel's traditional PC they center in network edge businesses and the rest from advanced graphics HPC, Mobileye and Foundry. Okay, that sounds pretty good. But it has to be taken into context that the balance of the semiconductor industry, yeah, this would be a pretty competitive growth rate, in our view, especially for a 70 plus billion dollar company. So kudos to Pat for sticking his neck out on this one. But again, the promise is several years away, at least four years away. Now we want to focus on Foundry because that's the only way Intel is going to get back into the volume game and the volume necessary for the company to compete. Pat built this slide showing the baby blue for today's Foundry business just under a billion dollars and adding in another $1.5 billion for Tower Semiconductor, the Israeli firm that it just acquired. So a few billion dollars in the near term future for the Foundry business. And then by 2026, this really fuzzy blue bar. Now remember, TSM is the new volume leader, and is a $50 billion company growing. So there's definitely a market there that it can go after. And adding in ARM processors to the mix, and, you know, opening up and partnering with the ecosystems out there can only help volume if Intel can win that business, which you know, it should be able to, given the likelihood of long term supply constraints. But we remain skeptical. This is another chart Pat showed, which makes the case that Foundry and IDM 2.0 will allow expensive assets to have a longer useful life. Okay, that's cool. It will also solve the cumulative output problem highlighted in the bottom right. We've talked at length about Wright's Law. That is, for every cumulative doubling of units manufactured, cost will fall by a constant percentage. You know, let's say around 15% in semiconductor world, which is vitally important to accommodate next generation chips, which are always more expensive at the start of the cycle. So you need that 15% cost buffer to jump curves and make any money. So let's unpack this a bit. You know, does this chart at the bottom right address our Wright's Law concerns, i.e. that Intel can't take advantage of Wright's Law because it can't double cumulative output fast enough? Now note the decline in wafer starts and then the slight uptick, and then the flattening. It's hard to tell what years we're talking about here. Intel is not going to share the sausage making because it's probably not pretty, But you can see on the bottom left, the flattening of the cumulative output curve in IDM 1.0 otherwise known as the death spiral. Okay, back to the power of Wright's Law. Now, assume for a second that wafer density doesn't grow. It does, but just work with us for a second. Let's say you produce 50 million units per year, just making a number up. That gets you cumulative output to $100 million in, sorry, 100 million units in the second year to take you two years to get to that 100 million. So in other words, it takes two years to lower your manufacturing cost by, let's say, roughly 15%. Now, assuming you can get wafer volumes to be flat, which that chart showed, with good yields, you're at 150 now in year three, 200 in year four, 250 in year five, 300 in year six, now, that's four years before you can take advantage of Wright's Law. You keep going at that flat wafer start, and that simplifying assumption we made at the start and 50 million units a year, and well, you get to the point. You get the point, it's now eight years before you can get the Wright's Law to kick in, and you know, by then you're cooked. But now you can grow the density of transistors on a chip, right? Yes, of course. So let's come back to Moore's Law. The graphic on the left says that all the growth is in the new stuff. Totally agree with that. Huge term that Pat presented. Now he also said that until we exhaust the periodic table of elements, Moore's Law is alive and well, and Intel is the steward of Moore's Law. Okay, that's cool. The chart on the right shows Intel going from 100 billion transistors today to a trillion by 2030. Hold that thought. So Intel is assuming that we'll keep up with Moore's Law, meaning a doubling of transistors every let's say two years, and I believe it. So bring that back to Wright's Law, in the previous chart, it means with IDM 2.0, Intel can get back to enjoying the benefits of Wright's Law every two years, let's say, versus IDM 1.0 where they were failing to keep up. Okay, so Intel is saved, yeah? Well, let's bring into this discussion one of our favorite examples, Apple's M1 ARM-based chip. The M1 Ultra is a new architecture. And you can see the stats here, 114 billion transistors on a five nanometer process and all the other stats. The M1 Ultra has two chips. They're bonded together. And Apple put an interposer between the two chips. An interposer is a pathway that allows electrical signals to pass through it onto another chip. It's a super fast connection. You can see 2.5 terabytes per second. But the brilliance is the two chips act as a single chip. So you don't have to change the software at all. The way Intel's architecture works is it takes two different chips on a substrate, and then each has its own memory. The memory is not shared. Apple shares the memory for the CPU, the NPU, the GPU. All of it is shared, meaning it needs no change in software unlike Intel. Now Intel is working on a new architecture, but Apple and others are way ahead. Now let's make this really straightforward. The original Apple M1 had 16 billion transistors per chip. And you could see in that diagram, the recently launched M1 Ultra has $114 billion per chip. Now if you take into account the size of the chips, which are increasing, and the increase in the number of transistors per chip, that transistor density, that's a factor of around 6x growth in transistor density per chip in 18 months. Remember Intel, assuming the results in the two previous charts that we showed, assuming they were achievable, is running at 2x every two years, versus 6x for the competition. And AMD and Nvidia are close to that as well because they can take advantage of TSM's learning curve. So in the previous chart with Moore's Law, alive and well, Intel gets to a trillion transistors by 2030. The Apple ARM and Nvidia ecosystems will arrive at that point years ahead of Intel. That means lower costs and significantly better competitive advantage. Okay, so where does that leave Intel? The story is really not resonating with investors and hasn't for a while. On February 18th, the day after its investor meeting, the stock was off. It's rebound a little bit but investors are, you know, they're probably prudent to wait unless they have really a long term view. And you can see Intel's performance relative to some of the major competitors. You know, Pat talked about five nodes in for years. He made a big deal out of that, and he shared proof points with Alder Lake and Meteor Lake and other nodes, but Intel just delayed granite rapids last month that pushed it out from 2023 to 2024. And it told investors that we're going to have to boost spending to turn this ship around, which is absolutely the case. And that delay in chips I feel like the first disappointment won't be the last. But as we've said many times, it's very difficult, actually, it's impossible to quickly catch up in semiconductors, and Intel will never catch up without volume. So we'll leave you by iterating our scenario that could save Intel, and that's if its Foundry business can eventually win back Apple to supercharge its volume story. It's going to be tough to wrestle that business away from TSM especially as TSM is setting up shop in Arizona, with US manufacturing that's going to placate The US government. But look, maybe the government cuts a deal with Apple, says, hey, maybe we'll back off with the DOJ and FTC and as part of the CHIPS Act, you'll have to throw some business at Intel. Would that be enough when combined with other Foundry opportunities Intel could theoretically produce? Maybe. But from this vantage point, it's very unlikely Intel will gain back its true number one leadership position. If it were really paranoid back when David Floyer sounded the alarm 10 years ago, yeah, that might have made a pretty big difference. But honestly, the best we can hope for is Intel's strategy and execution allows it to get competitive volumes by the end of the decade, and this national treasure survives to fight for its leadership position in the 2030s. Because it would take a miracle for that to happen in the 2020s. Okay, that's it for today. Thanks to David Floyer for his contributions to this research. Always a pleasure working with David. Stephanie Chan helps me do much of the background research for "Breaking Analysis," and works with our CUBE editorial team. Kristen Martin and Cheryl Knight to get the word out. And thanks to SiliconANGLE's editor in chief Rob Hof, who comes up with a lot of the great titles that we have for "Breaking Analysis" and gets the word out to the SiliconANGLE audience. Thanks, guys. Great teamwork. Remember, these episodes are all available as podcast wherever you listen. Just search "Breaking Analysis Podcast." You'll want to check out ETR's website @etr.ai. We also publish a full report every week on wikibon.com and siliconangle.com. You could always get in touch with me on email, david.vellante@siliconangle.com or DM me @dvellante, and comment on my LinkedIn posts. This is Dave Vellante for "theCUBE Insights, Powered by ETR." Have a great week. Stay safe, be well, and we'll see you next time. (upbeat music)

>> From theCUBE studios in Palo Alto and Boston, bringing you data-driven insights from theCUBE and ETR. This is Breaking Analysis with Dave Vellante. >> Techniques to protect sensitive data have evolved over thousands of years literally. The pace of modern data protection is rapidly accelerating and presents both opportunities and threats for organizations. In particular, the amount of data stored in the cloud combined with hybrid work models, the clear and present threat of cyber crime, regulatory edicts and the ever expanding edge and associated use cases should put CXOs on notice that the time is now to rethink your data protection strategies. Hello, and welcome to this week's Wikibon theCUBE Insights powered by ETR. In this Breaking Analysis, we're going to explore the evolving world of data protection and share some information on how we see the market changing in the competitive landscape for some of the top players. Steve Kenniston AKA the Storage Alchemist shared a story with me and it was pretty clever. Way back in 4,000 BC the Sumerians invented the first system of writing. Now they used clay tokens to represent transactions at that time. Now, to prevent messing with these tokens, they sealed them in clay jars to ensure that the tokens or either data would remain secure with an accurate record, let's call it quasi immutable and lived in a clay vault. Since that time, we've seen quite an evolution in data protection. Tape, of course, was the main means of protecting data, backing data up during most of the mainframe era and that carried into client server computing, which really accentuated and underscored the issues around backup windows and challenges with RTO, Recovery Time Objective and RPO, Recovery Point Objective, and just overall recovery nightmares. Then in the 2000s data reduction made displace backup more popular and push tape into an archive last resort media data domain then EMC now Dell still sell many purpose built backup appliances as do others as a primary backup target disc base. The rise of virtualization brought more changes in backup and recovery strategies as a reduction in physical resources squeezed the one application that wasn't under utilizing compute i.e backup. And we saw the rise of Veeam, the cleverly named company that became synonymous with data protection for virtual machines. Now the cloud has created new challenges related to data sovereignty, governance latency, copy creep, expense, et cetera but more recently cyber threats have elevated data protection to become a critical adjacency to information security. Cyber resilience to specifically protect against ransomware attacks as the new trend being pushed by the vendor community as organizations are urgently looking for help with this insidious threat. Okay, so there are two major disruptors that we're going to talk about today, the cloud and cyber crime, especially around ransoming your data. Every customer is using the cloud in some way, shape or form. Around 76% are using multiple clouds that's according to a recent study by HashiCorp. We've talked extensively about skill shortages on theCUBE and data protection and security concerns are really key challenges to address given that skill shortage is a real talent gap in terms of being able to throw people at solving this problem. So what customers are doing they're either building out or they're buying, really mostly building abstraction layers to hide the underlying cloud complexity. So, what this does, the good news is it simplifies provisioning and management but it creates problems around opacity. In other words, you can't see sometimes what's going on with the data, these challenges fundamentally become data problems in our view. Things like fast, accurate, and complete backup recovery, compliance, data sovereignty, data sharing, I mentioned copy creep, cyber resiliency, privacy protections these are all challenges brought to fore by the cloud, the advantages, the pros and the cons. Now, remote workers are especially vulnerable and as clouds expand rapidly data protection technologies are struggling to keep pace. So let's talk briefly about the rapidly expanding public cloud. This chart shows worldwide revenue for the big four hyperscalers, as you can see we projected they're going to surpass $115 billion in revenue in 2021, that's up from 86 billion last year. So it's a huge market, it's growing in the 35% range. The interesting thing is last year, 80 plus billion dollars in revenue but a 100 billion dollars was spent last year by these firms in CapEx. So they're building out infrastructure for the industry. This is a gift to the balance of the industry. Now to date legacy vendors and their surrounding community have been pretty defensive around the cloud, "Oh, not everything is going to move to the cloud, it's not a zero sum game we here." And while that's all true the narrative was really kind of a defense posture and that's starting to change as large tech companies like Dell, IBM, Cisco, HPE, and others see opportunities to build on top of this infrastructure. You certainly see that with Arvind Krishna's comments at IBM, Cisco obviously leaning in from a networking and security perspective. HPE using language that is very much cloud-like with its GreenLake strategy. And of course, Dell is all over this. Let's listen to how Michael Dell is thinking about this opportunity when he was questioned on theCUBE by John Furrier about the cloud. Play the clip. >> Well, clouds are infrastructure, right? So you can have a public cloud, you can have an edge cloud, a private cloud, a Telco cloud, a hybrid cloud, multicloud, here cloud, there cloud, everywhere cloud, cloud. Yet, they'll all be there, but it's basically infrastructure. And how do you make that as easy to consume and create the flexibility that enables everything. >> Okay, so in my view, Michael nailed it, the cloud is everywhere. You have to make it easy and you have to admire the scope of his comments. We know this guy, he thinks big, right? He said enables everything. What he's basically saying is that, technology is at the point where it has the potential to touch virtually every industry, every person, every problem, everything. So let's talk about how this informs the changing world of data protection. Now, we've seen with the pandemic there's an acceleration toward digital and that has caused an escalation if you will, in the data protection mandate. So essentially what we're talking about here is the application of Michael Dell's cloud everywhere comments. You've got on-prem, private clouds, hybrid clouds, you've got public clouds across AWS, Azure, Google, Alibaba, really those big four hyperscalers. You got many clouds that are popping up all over the place, but multicloud to that HashiCorp data point, 75, 76%, and then you now see the cloud expanding out to the edge, programmable infrastructure heading out to the edge. So the opportunity here to build the data protection cloud is to have the same experiences across all these estates with automation and orchestration in that cloud, that data protection cloud if you will. So think of it as an abstraction layer that hides that underlying complexity, you log into that data protection cloud it's the same experience. So you've got backup, you've got recovery, you can handle bare-metal, you can do virtualized backups and recoveries, any cloud, any OS, out to the edge, Kubernetes and container use cases, which is an emerging data protection requirement and you've got analytics, perhaps you've got PII, Personally Identifiable Information protection in there. So the attributes of this data protection cloud, again, it abstracts the underlying cloud primitives, takes care of that. It also explodes cloud native technologies. In other words, it takes advantage of whether it's machine learning, which all the big cloud players have expertise in, new processor models things like Graviton and other services that are in the cloud natively. It doesn't just wrap it's on-prem stack in a container and shove it into the cloud, no, it actually re architects or architects around those cloud native services and it's got distributed metadata to track files and volumes and any organizational data irrespective of location. And it enables sets of services to intelligently govern in a federated governance manner while ensuring data integrity and all this is automated and orchestrated to help with the skills gap. Now, as it relates to cyber recovery, air gap solutions must be part of the portfolio, but managed outside of that data protection cloud that we just briefly described. The orchestration and the management must also be gapped if you will, otherwise, you don't have an air gap. So all of this is really a cohort to cyber security or your cybersecurity strategy and posture, but you have to be careful here because your data protection strategy could get lost in this mess. So you want to think about the data protection cloud as again, an adjacency or maybe an overlay to your cybersecurity approach, not a bolt on it's got to be fundamentally architectured from the bottom up. And yes, this is going to maybe create some overheads and some integration challenges but this is the way in which we think you should think about it. So you'll likely need a partner to do this, again, we come back to the skills gap if were seeing the rise of MSPs, managed service providers and specialist service providers, not public cloud providers, people are concerned about lock-in and that's really not their role. They're not high touch services company, probably not your technology arms dealer, excuse me, they're selling technology to these MSPs. So the MSPs, they have intimate relationships with their customers. They understand their business and specialize in architecting solutions to handle these difficult challenges. So let's take a look at some of the risk factors here and dig a little bit into the cyber threat that organizations face. This is a slide that, again, the Storage Alchemists, Steve Kenniston shared with me, it's based on a study that IBM funds with the Panama Institute, which is a firm that studies these things like cost of breaches and has for many, many, many years. The slide shows the total cost of a typical breach within each dot and on the Y-axis and the frequency in percentage terms on the horizontal axis. Now it's interesting, the top two are compromised credentials and fishing, which once again proves that bad user behavior trumps good security every time. But the point here is that the adversary's attack vectors are many and specific companies often specialize in solving these problems often with point products, which is why the slide that we showed from Optiv earlier, that messy slide looks so cluttered. So it's a huge challenge for companies, and that's why we've seen the emergence of cyber recovery solutions from virtually all the major players. Ransomware and the SolarWinds hack have made trust the number one issue for CEOs and CSOs and boards of directors, shifting CSO spending patterns are clear. Shifting largely because they're catalyzed by the work from home. But outside of the moat to endpoint security identity and access management, cloud security, the horizontal network security. So security priorities and spending are changing that's why you see the emergence of disruptors like we've covered extensively, Okta, Crowdstrike, Zscaler. And cyber resilience is top of mind and robust solutions are required and that's why companies are building cyber recovery solutions that are most often focused on the backup corpus because that's a target for the bad guys. So there is an opportunity, however to expand from just the backup corpus to all data and protect this kind of 3-2-1, or maybe it's 3-2-1-1, three copies, two backups, a backup in the cloud and one that's air gapped. So this can be extended to primary storage, copies, snaps, containers, data in motion, et cetera, to have a comprehensive data protection strategy. Customers as I said earlier, increasingly looking to manage service providers and specialists because of that skills gap and that's a big reason why automation is so important in orchestration. And automation and orchestration I'll emphasize on the air gap solutions should be separated physically and logically. All right, now let's take a look at some of the ETR data and some of the players. This is a chart that we like to show often, it's a X, Y axis, and the Y-axis is net score, which is a measure of spending momentum and the horizontal axis is market share. Now market share is an indicator of pervasiveness in the survey. It's not spending market share, it's not market share of the overall market, it's a term that ETR uses. It's essentially market share of the responses within the survey set, think of it as mind share. Okay, you've got the pure plays here on this slide in the storage category, there is no data protection or backup category so what we've done is we've isolated the pure plays or close to pure plays in backup and data protection. Notice that red line, that red line is kind of our subjective view of anything that's over that 40% line is elevated, you can see only rubric in the July survey is over that 40% line. I'll show you the ends in a moment. Smaller ends, but still rubric is the only one. Now look at Cohesity and rubric in the January, 2020. So last year pre-pandemic Cohesity and Rubrik they've come well off their peaks for net score. Look at Veeam, Veeam having studied this data for the last say 24 plus months, Veeam has been Steady Eddie. It is really always in the mid to high 30s, always shows a large shared end so it's coming up in the survey, customers are mentioning Veeam and it's got a very solid net score. It's not above that 40% line but it's hovering just below consistently, that's very impressive. Commvault has steadily been moving up. Sanjay Mirchandani has made some acquisitions, he did the Hedvig acquisition. They launched metallic that's driving cloud affinity within a Commvault large customer base so it's a good example of a legacy player, pivoting and evolving and transforming itself. Veritas continues to underperform in the ETR surveys relative to the other players. Now, for context, let's say add IBM and Dell to the chart. Now just note, this is IBM and Dell's full storage portfolio. The category in the taxonomy at ETR is all storage. Okay, this previous slide I isolated on the pure plays, but this now adds in IBM and Dell. It probably representative of where they would be, probably Dell larger on the horizontal axis than IBM, of course and you could see the spending momentum in accordingly. So you could see that in the data chart that we've inserted. So smaller ends for Rubrik and Cohesity, but still enough to pay attention, it's not like one or two when you're 20 plus, 15 plus, 25 plus you can start to pay attention to trends. Veeam again is very impressive. Its net score is solid, it's got a consistent presence in the dataset, it's clear leader here. SimpliVity is small but it's improving relative to last several surveys and we talked about Commvault. Now, I want to emphasize something that we've been hitting on for quite some time now and that's the renaissance that's coming in compute. Now we all know about Moore's law, the doubling of transistor density every two years, 18 to 24 months and that leads to a doubling of performance in that time frame. X86, that X86 curve is in the blue and if you do the math, this is expressed in trillions of operations per second. The orange line is a representative of Apple's A series culminating in the A-15 most recently, the A series is what Apple is now... It's the technology basis for what's inside, and one the new Apple laptops, which is replacing Intel. That's that orange line there we'll come back to that. So go back to the blue line for a minute. If you do the math on doubling performance every 24 months, it comes out to roughly 40% annual improvement in processing power per year. That's now moderated. So Moore's law is waning in one sense so we wrote a piece Moore's law is not dead so I'm sort of contradicting myself there, but the traditional Moore's law curve on X86 is waning. It's probably now down to around 30%, low 30s, but look at the orange line. Again, using the A series as an indicator, if you combine the CPU, the NPU, which is the neural processing unit, XPU, pick whatever PU you want, the accelerators, the DSPs, that line is growing at a 100% plus per year. It's probably more accurately around 110% a year. So there's a new industry curve occurring and it's being led by the Arm ecosystem. The other key factor there you see in a lot of use cases, a lot of consumer use cases Apple is an example but you're also seeing it in things like Tesla, Amazon with AWS Graviton, the Annapurna acquisition, building out Graviton and Nitro that's based on Arm. You can get from design to tape out in less than two years Whereas the Intel cycles we know they've been running it four to five years now, maybe Pat Gelsinger is compressing those, but Intel is behind. So, organizations that are on that orange curve are going to see faster acceleration, lower cost, lower power, et cetera. All right, so what's the tie to data protection? I'm going to leave you with this chart. Arm has introduced it's confidential compute architecture, and is ushering in a new era of security and data protection. Zero Trust is the new mandate and what Arm has done with what they call realms is create physical separation of the vulnerable components by creating essentially physical buckets to put code in and to put data in separate from the OS. Remember the OS is the most valuable entry point for hackers or one of them because it contains privileged access and it's a weak link because of things like memory leakages and vulnerabilities. And malicious code can be placed by bad guys within data in the OS and appear benign even though it's anything but. So in this architecture, all the OS does is create API calls to the realm controller. That's the only interaction. So it makes it much harder for bad actors to get access to the code and the data. And importantly, very importantly, it's an end-to-end architecture so there's protection throughout if you're pulling data from the edge and bringing it back to on-prem and the cloud you've got that end-to-end architecture and protection throughout. So the link to data protection is that backup software vendors need to be the most trusted of applications. Backup software needs to be the most trusted of applications because it's one of the most targeted areas in the cyber attack. Realms provide an end-to-end separation of data and code from the OS and is a better architectural construct to support Zero Trust and confidential computing and critical use cases like data protection/backup and other digital business apps. So our call to action is backup software vendors you can lead the charge. Arm is several years ahead at the moment, head of Intel in our view. So you got to pay attention to that, research that, we're not saying over rotate, but go investigate that. And use your relationships with Intel to accelerate its version of this architecture or ideally the industry should agree on common standards and solve this problem together. Pat Gelsinger told us in theCUBE that if it's the last thing he's going to do in his industry life he's going to solve this security problem. That's when he was at VMware. Well, Pat you're even in a better place to do it now, you don't have to solve it yourself, you can't and you know that. So while you're going about your business saving Intel, look to partner with Arm I know it sounds crazy to use these published APIs and push to collaborate on an open source architecture that addresses the cyber problem. If anyone can do it, you can. Okay, that's it for today. Remember, these episodes are all available as podcasts all you got to do is search Breaking Analysis podcast, I publish weekly on Wikibon.com and SiliconANGLE.com. Or you can reach me at dvellante on Twitter, email me at Dave.Vellante@SiliconANGLE.com. And don't forget to check out ETR.plus for all the survey and data action. This is Dave Vellante for theCUBE Insights powered by ETR. Thanks for watching everybody, be well and we'll see you next time. (upbeat music)

>> From theCUBE studios in Palo Alto in Boston, bringing you data-driven insights from theCUBE and ETR. This is braking analysis with Dave Vellante. >> Techniques to protect sensitive data have evolved over thousands of years, literally. The pace of modern data protection is rapidly accelerating and presents both opportunities and threats for organizations. In particular, the amount of data stored in the cloud combined with hybrid work models, the clear and present threat of cyber crime, regulatory edicts, and the ever expanding edge and associated use cases should put CXOs on notice that the time is now to rethink your data protection strategies. Hello, and welcome to this week's Wikibon Cube Insights powered by ETR. In this breaking analysis, we're going to explore the evolving world of data protection and share some information on how we see the market changing in the competitive landscape for some of the top players. Steve Kenniston, AKA the Storage Alchemist, shared a story with me, and it was pretty clever. Way back in 4000 BC, the Sumerians invented the first system of writing. Now, they used clay tokens to represent transactions at that time. Now, to prevent messing with these tokens, they sealed them in clay jars to ensure that the tokens, i.e the data, would remain secure with an accurate record that was, let's call it quasi, immutable, and lived in a clay vault. And since that time, we've seen quite an evolution of data protection. Tape, of course, was the main means of protecting data and backing data up during most of the mainframe era. And that carried into client server computing, which really accentuated and underscored the issues around backup windows and challenges with RTO, recovery time objective and RPO recovery point objective. And just overall recovery nightmares. Then in the 2000's data reduction made disk-based backup more popular and pushed tape into an archive last resort media. Data Domain, then EMC, now Dell still sell many purpose-built backup appliances as do others as a primary backup target disc-based. The rise of virtualization brought more changes in backup and recovery strategies, as a reduction in physical resources squeezed the one application that wasn't under utilizing compute, i.e, backup. And we saw the rise of Veem, the cleverly-named company that became synonymous with data protection for virtual machines. Now, the cloud has created new challenges related to data sovereignty, governance, latency, copy creep, expense, et cetera. But more recently, cyber threats have elevated data protection to become a critical adjacency to information security. Cyber resilience to specifically protect against attacks is the new trend being pushed by the vendor community as organizations are urgently looking for help with this insidious threat. Okay, so there are two major disruptors that we're going to talk about today, the cloud and cyber crime, especially around ransoming your data. Every customer is using the cloud in some way, shape, or form. Around 76% are using multiple clouds, that's according to a recent study by Hashi Corp. We've talked extensively about skill shortages on theCUBE, and data protection and security concerns are really key challenges to address, given that skill shortage is a real talent gap in terms of being able to throw people at solving this problem. So what customers are doing, they're either building out or they're buying really mostly building abstraction layers to hide the underlying cloud complexity. So what this does... The good news is it's simplifies provisioning and management, but it creates problems around opacity. In other words, you can't see sometimes what's going on with the data. These challenges fundamentally become data problems, in our view. Things like fast, accurate, and complete backup recovery, compliance, data sovereignty, data sharing. I mentioned copy creep, cyber resiliency, privacy protections. These are all challenges brought to fore by the cloud, the advantages, the pros, and the cons. Now, remote workers are especially vulnerable. And as clouds span rapidly, data protection technologies are struggling to keep pace. So let's talk briefly about the rapidly-expanding public cloud. This chart shows worldwide revenue for the big four hyperscalers. As you can see, we projected that they're going to surpass $115 billion in revenue in 2021. That's up from 86 billion last year. So it's a huge market, it's growing in the 35% range. The interesting thing is last year, 80-plus billion dollars in revenue, but 100 billion dollars was spent last year by these firms in cap ex. So they're building out infrastructure for the industry. This is a gift to the balance of the industry. Now to date, legacy vendors and the surrounding community have been pretty defensive around the cloud. Oh, not everything's going to move to the cloud. It's not a zero sum game we hear. And while that's all true, the narrative was really kind of a defensive posture, and that's starting to change as large tech companies like Dell, IBM, Cisco, HPE, and others see opportunities to build on top of this infrastructure. You certainly see that with Arvind Krishna comments at IBM, Cisco obviously leaning in from a networking and security perspective, HPE using language that is very much cloud-like with its GreenLake strategy. And of course, Dell is all over this. Let's listen to how Michael Dell is thinking about this opportunity when he was questioned on the queue by John Furrier about the cloud. Play the clip. So in my view, Michael nailed it. The cloud is everywhere. You have to make it easy. And you have to admire the scope of his comments. We know this guy, he thinks big. He said, "Enables everything." He's basically saying is that technology is at the point where it has the potential to touch virtually every industry, every person, every problem, everything. So let's talk about how this informs the changing world of data protection. Now, we all know, we've seen with the pandemic, there's an acceleration in toward digital, and that has caused an escalation, if you will, in the data protection mandate. So essentially what we're talking about here is the application of Michael Dell's cloud everywhere comments. You've got on-prem, private clouds, hybrid clouds. You've got public clouds across AWS, Azure, Google, Alibaba. Really those are the big four hyperscalers. You got many clouds that are popping up all their place. But multi-cloud, to that Hashi Corp data point, 75, 70 6%. And then you now see the cloud expanding out to the edge, programmable infrastructure heading out to the edge. So the opportunity here to build the data protection cloud is to have the same experiences across all these estates with automation and orchestration in that cloud, that data protection cloud, if you will. So think of it as an abstraction layer that hides that underlying complexity, you log into that data protection cloud, it's the same experience. So you've got backup, you've got recovery, you can handle bare metal. You can do virtualized backups and recoveries, any cloud, any OS, out to the edge, Kubernetes and container use cases, which is an emerging data protection requirement. And you've got analytics, perhaps you've got PII, personally identifiable information protection in there. So the attributes of this data protection cloud, again, abstracts the underlying cloud primitives, takes care of that. It also explodes cloud native technologies. In other words, it takes advantage of whether it's machine learning, which all the big cloud players have expertise in, new processor models, things like graviton, and other services that are in the cloud natively. It doesn't just wrap it's on-prem stack in a container and shove it into the cloud, no. It actually re architects or architects around those cloud native services. And it's got distributed metadata to track files and volumes and any organizational data irrespective of location. And it enables sets of services to intelligently govern in a federated governance manner while ensuring data integrity. And all this is automated and an orchestrated to help with the skills gap. Now, as it relates to cyber recovery, air-gap solutions must be part of the portfolio, but managed outside of that data protection cloud that we just briefly described. The orchestration and the management must also be gaped, if you will. Otherwise, (laughs) you don't have an air gap. So all of this is really a cohort to cyber security or your cybersecurity strategy and posture, but you have to be careful here because your data protection strategy could get lost in this mess. So you want to think about the data protection cloud as again, an adjacency or maybe an overlay to your cybersecurity approach. Not a bolt on, it's got to be fundamentally architectured from the bottom up. And yes, this is going to maybe create some overheads and some integration challenges, but this is the way in which we think you should think about it. So you'll likely need a partner to do this. Again, we come back to the skill skills gap if we're seeing the rise of MSPs, managed service providers and specialist service providers. Not public cloud providers. People are concerned about lock-in, and that's really not their role. They're not high-touch services company. Probably not your technology arms dealer, (clear throat) excuse me, they're selling technology to these MSPs. So the MSPs, they have intimate relationships with their customers. They understand their business and specialize in architecting solutions to handle these difficult challenges. So let's take a look at some of the risk factors here, dig a little bit into the cyber threat that organizations face. This is a slide that, again, the Storage Alchemists, Steve Kenniston, shared with me. It's based on a study that IBM funds with the Panmore Institute, which is a firm that studies these things like cost of breaches and has for many, many, many years. The slide shows the total cost of a typical breach within each dot and on the Y axis and the frequency in percentage terms on the horizontal axis. Now, it's interesting. The top two compromise credentials and phishing, which once again proves that bad user behavior trumps good security every time. But the point here is that the adversary's attack vectors are many. And specific companies often specialize in solving these problems often with point products, which is why the slide that we showed from Optiv earlier, that messy slide, looks so cluttered. So there's a huge challenge for companies. And that's why we've seen the emergence of cyber recovery solutions from virtually all the major players. Ransomware and the solar winds hack have made trust the number one issue for CIOs and CISOs and boards of directors. Shifting CISO spending patterns are clear. They're shifting largely because they're catalyzed by the work from home. But outside of the moat to endpoint security, identity and access management, cloud security, the horizontal network security. So security priorities and spending are changing. And that's why you see the emergence of disruptors like we've covered extensively, Okta, CrowdStrike, Zscaler. And cyber resilience is top of mind, and robust solutions are required. And that's why companies are building cyber recovery solutions that are most often focused on the backup corpus because that's a target for the bad guys. So there is an opportunity, however, to expand from just the backup corpus to all data and protect this kind of 3, 2, 1, or maybe it's 3, 2, 1, 1, three copies, two backups, a backup in the cloud and one that's air gaped. So this can be extended to primary storage, copies, snaps, containers, data in motion, et cetera, to have a comprehensive data protection strategy. And customers, as I said earlier, are increasingly looking to manage service providers and specialists because of that skills gap. And that's a big reason why automation is so important in orchestration. And automation and orchestration, I'll emphasize, on the air gap solutions should be separated physically and logically. All right, now let's take a look at some of the ETR data and some of the players. This is a chart that we like to show often. It's a X-Y axis. And the Y axis is net score, which is a measure of spending momentum. And the horizontal axis is market share. Now, market share is an indicator of pervasiveness in the survey. It's not spending market share, it's not market share of the overall market, it's a term that ETR uses. It's essentially market share of the responses within the survey set. Think of it as mind share. Okay, you've got the pure plays here on this slide, in the storage category. There is no data protection or backup category. So what we've done is we've isolated the pure plays or close to pure plays in backup and data protection. Now notice that red line, that red is kind of our subjective view of anything that's over that 40% line is elevated. And you can see only Rubrik, and the July survey is over that 40% line. I'll show you the ends in a moment. Smaller ends, but still, Rubrik is the only one. Now, look at Cohesity and Rubrik in the January 2020. So last year, pre-pandemic, Cohesity and Rubrik, they've come well off their peak for net score. Look at Veeam. Veeam, having studied this data for the last say 24 hours months, Veeam has been steady Eddy. It is really always in the mid to high 30s, always shows a large shared end, so it's coming up in the survey. Customers are mentioning Veeam. And it's got a very solid net score. It's not above that 40% line, but it's hovering just below consistently. That's very impressive. Commvault has steadily been moving up. Sanjay Mirchandani has made some acquisitions. He did the Hedvig acquisition. They launched Metallic, that's driving cloud affinity within Commvault's large customer base. So it's good example of a legacy player pivoting and evolving and transforming itself. Veritas, it continues to under perform in the ETR surveys relative to the other players. Now, for context, let's add IBM and Dell to the chart. Now just note, this is IBM and Dell's full storage portfolio. The category in the taxonomy at ETR is all storage. Just previous slide, I isolated on the pure plays. But this now adds in IBM and Dell. It probably representative of where they would be. Probably Dell larger on the horizontal axis than IBM, of course. And you could see the spending momentum accordingly. So you can see that in the data chart that we've inserted. So some smaller ends for Rubrik and Cohesity. But still enough to pay attention, it's not like one or two. When you're 20-plus, 15-plus 25-plus, you can start to pay attention to trends. Veeam, again, is very impressive. It's net score is solid, it's got a consistent presence in the dataset, it's clear leader here. SimpliVity is small, but it's improving relative to last several surveys. And we talked about Convolt. Now, I want to emphasize something that we've been hitting on for quite some time now. And that's the Renaissance that's coming in compute. Now, we all know about Moore's Law, the doubling of transistor density every two years, 18 to 24 months. And that leads to a doubling of performance in that timeframe. X86, that x86 curve is in the blue. And if you do the math, this is expressed in trillions of operations per second. The orange line is representative of Apples A series, culminating in the A15, most recently. The A series is what Apple is now... Well, it's the technology basis for what's inside M1, the new Apple laptops, which is replacing Intel. That's that that orange line there, we'll come back to that. So go back to the blue line for a minute. If you do the math on doubling performance every 24 months, it comes out to roughly 40% annual improvement in processing power per year. That's now moderated. So Moore's Law is waning in one sense, so we wrote a piece Moore's Law is not dead. So I'm sort of contradicting myself there. But the traditional Moore's Law curve on x86 is waning. It's probably now down to around 30%, low 30s. But look at the orange line. Again, using the A series as an indicator, if you combine then the CPU, the NPU, which neuro processing unit, XPU, pick whatever PU you want, the accelerators, the DSPs, that line is growing at 100% plus per year. It's probably more accurately around 110% a year. So there's a new industry curve occurring, and it's being led by the Arm ecosystem. The other key factor there, and you're seeing this in a lot of use cases, a lot of consumer use cases, Apple is an example, but you're also seeing it in things like Tesla, Amazon with AWS graviton, the Annapurna acquisition, building out graviton and nitro, that's based on Arm. You can get from design to tape out in less than two years. Whereas the Intel cycles, we know, they've been running it four to five years now. Maybe Pat Gelsinger is compressing those. But Intel is behind. So organizations that are on that orange curve are going to see faster acceleration, lower cost, lower power, et cetera. All right, so what's the tie to data protection. I'm going to leave you with this chart. Arm has introduced it's confidential, compute architecture and is ushering in a new era of security and data protection. Zero trust is the new mandate. And what Arm has it's done with what they call realms is create physical separation of the vulnerable components by creating essentially physical buckets to put code in and to put data in, separate from the OS. Remember, the OS is the most valuable entry point for hackers or one of them because it contains privileged access, and it's a weak link because of things like memory leakages and vulnerabilities. And malicious code can be placed by bad guys within data in the OS and appear benign, even though it's anything but. So in this, all the OS does is create API calls to the realm controller. That's the only interaction. So it makes it much harder for bad actors to get access to the code and the data. And importantly, very importantly, it's an end-to-end architecture. So there's protection throughout. If you're pulling data from the edge and bringing it back to the on-prem or the cloud, you've got that end to end architecture and protection throughout. So the link to data protection is that backup software vendors need to be the most trusted of applications. Backup software needs to be the most trusted of applications because it's one of the most targeted areas in a cyber attack. Realms provide an end-to-end separation of data and code from the OS and it's a better architectural construct to support zero trust and confidential computing and critical use cases like data protection/backup and other digital business apps. So our call to action is backup software vendors, you can lead the charge. Arm is several years ahead at the moment, ahead of Intel, in our view. So you've got to pay attention to that, research that. We're not saying over rotate, but go investigate that. And use your relationships with Intel to accelerate its version of this architecture. Or ideally, the industry should agree on common standards and solve this problem together. Pat Gelsinger told us in theCUBE that if it's the last thing he's going to do in his industry life, he's going to solve this security problem. That's when he was at VMware. Well, Pat, you're even in a better place to do it now. You don't have to solve it yourself, you can't, and you know that. So while you're going about your business saving Intel, look to partner with Arm. I know it sounds crazy to use these published APIs and push to collaborate on an open source architecture that addresses the cyber problem. If anyone can do it, you can. Okay, that's it for today. Remember, these episodes are all available as podcasts. All you got to do is search Braking Analysis Podcast. I publish weekly on wikibond.com and siliconangle.com. Or you can reach me @dvellante on Twitter, email me at david.vellante@siliconangle.com. And don't forget to check out etr.plus for all the survey and data action. This is Dave Vellante for theCUBE Insights, powered by ETR. Thanks for watching, everybody. Be well, and we'll see you next time. (gentle music)

(upbeat music) >> Welcome back to HPE Discover 2021, theCube's virtual coverage, continuous coverage of HPE's annual customer event. My name is Dave Vellante and we're going to dive into the intersection of high-performance computing, data and AI with Dr. Eng Lim Goh who's a Senior Vice President and CTO for AI at Hewlett Packard Enterprise. Dr. Goh, great to see you again. Welcome back to theCube. >> Hey, hello, Dave. Great to talk to you again. >> You might remember last year we talked a lot about swarm intelligence and how AI is evolving. Of course you hosted the Day 2 keynotes here at Discover. And you talked about thriving in the age of insights and how to craft a data-centric strategy and you addressed some of the biggest problems I think organizations face with data. And that's, you got to look, data is plentiful, but insights, they're harder to come by and you really dug into some great examples in retail, banking, and medicine and healthcare and media. But stepping back a little bit we'll zoom out on Discover '21, you know, what do you make of the events so far and some of your big takeaways? >> Hmm, well, you started with the insightful question. Data is everywhere then but we lack the insight. That's also part of the reason why that's a main reason why, Antonio on Day 1 focused and talked about that, the fact that we are in the now in the age of insight and how to thrive in this new age. What I then did on the Day 2 keynote following Antonio is to talk about the challenges that we need to overcome in order to thrive in this new age. >> So maybe we could talk a little bit about some of the things that you took away in terms of, I'm specifically interested in some of the barriers to achieving insights when you know customers are drowning in data. What do you hear from customers? What were your takeaway from some of the ones you talked about today? >> Very pertinent question, Dave. You know, the two challenges I spoke about how to, that we need to overcome in order to thrive in this new age, the first one is the current challenge. And that current challenge is, you know state of this, you know, barriers to insight, when we are awash with data. So that's a statement. How to overcome those barriers. One of the barriers to insight when we are awash in data, in the Day 2 keynote, I spoke about three main things, three main areas that receive from customers. The first one, the first barrier is with many of our customers, data is siloed. You know, like in a big corporation, you've got data siloed by sales, finance, engineering, manufacturing, and so on supply chain and so on. And there's a major effort ongoing in many corporations to build a Federation layer above all those silos so that when you build applications above they can be more intelligent. They can have access to all the different silos of data to get better intelligence and more intelligent applications built. So that was the first barrier we spoke about, you know, barriers to insight when we are awash with data. The second barrier is that we see amongst our customers is that data is raw and disperse when they are stored. And it's tough to get to value out of them. In that case I use the example of the May 6, 2010 event where the stock market dropped a trillion dollars in tens of minutes. We all know those who are financially attuned with, know about this incident. But that this is not the only incident. There are many of them out there. And for that particular May 6, event, you know it took a long time to get insight, months, yeah, before we, for months we had no insight as to what happened, why it happened. And there were many other incidences like this and the regulators were looking for that one rule that could mitigate many of these incidences. One of our customers decided to take the hard road to go with the tough data. Because data is raw and dispersed. So they went into all the different feeds of financial transaction information, took the tough, you know, took a tough road and analyze that data took a long time to assemble. And he discovered that there was quote stuffing. That people were sending a lot of trades in and then canceling them almost immediately. You have to manipulate the market. And why didn't we see it immediately? Well, the reason is the process reports that everybody sees had the rule in there that says all trades less than 100 shares don't need to report in there. And so what people did was sending a lot of less than 100 shares trades to fly under the radar to do this manipulation. So here is, here the second barrier. Data could be raw and disperse. Sometimes it's just have to take the hard road and to get insight. And this is one great example. And then the last barrier has to do with sometimes when you start a project to get insight, to get answers and insight, you realize that all the data's around you, but you don't seem to find the right ones to get what you need. You don't seem to get the right ones, yeah. Here we have three quick examples of customers. One was a great example where they were trying to build a language translator a machine language translator between two languages. But in order to do that they need to get hundreds of millions of word pairs of one language compare with the corresponding other hundreds of millions of them. They say, "Where I'm going to get all these word pairs?" Someone creative thought of a willing source and huge source, it was a United Nations. You see, so sometimes you think you don't have the right data with you, but there might be another source and a willing one that could give you that data. The second one has to do with, there was the, sometimes you may just have to generate that data. Interesting one. We had an autonomous car customer that collects all these data from their cars. Massive amounts of data, lots of sensors, collect lots of data. And, you know, but sometimes they don't have the data they need even after collection. For example, they may have collected the data with a car in fine weather and collected the car driving on this highway in rain and also in snow. But never had the opportunity to collect the car in hail because that's a rare occurrence. So instead of waiting for a time where the car can drive in hail, they build a simulation by having the car collected in snow and simulated hail. So these are some of the examples where we have customers working to overcome barriers. You have barriers that is associated with the fact, that data silo, if federated barriers associated with data that's tough to get at. They just took the hard road. And sometimes thirdly, you just have to be creative to get the right data you need. >> Wow, I tell you, I have about 100 questions based on what you just said. And as a great example, the flash crash in fact Michael Lewis wrote about this in his book, the "Flash Boys" and essentially. It was high frequency traders trying to front run the market and sending in small block trades trying to get sort of front ended. So that's, and they chalked it up to a glitch. Like you said, for months, nobody really knew what it was. So technology got us into this problem. Can I guess my question is can technology help us get get out of the problem? And that maybe is where AI fits in. >> Yes. Yes. In fact, a lot of analytics work went in to go back to the raw data that is highly dispersed from different sources, assemble them to see if you can find a material trend. You can see lots of trends. Like, no, we, if humans at things we tend to see patterns in clouds. So sometimes you need to apply statistical analysis, math to be sure that what the model is seeing is real. And that required work. That's one area. The second area is, you know, when this, there are times when you just need to go through that tough approach to find the answer. Now, the issue comes to mind now is that humans put in the rules to decide what goes into a report that everybody sees. And in this case before the change in the rules. By the way, after the discovery, the authorities changed the rules and all shares all trades of different, any sizes it has to be reported. Not, yeah. But the rule was applied to to say earlier that shares under 100, trades under 100 shares need not be reported. So sometimes you just have to understand that reports were decided by humans and for understandable reasons. I mean, they probably didn't, wanted for various reasons not to put everything in there so that people could still read it in a reasonable amount of time. But we need to understand that rules were being put in by humans for the reports we read. And as such there are times we just need to go back to the raw data. >> I want to ask you-- Or be it that it's going to be tough there. >> Yeah, so I want to ask you a question about AI as obviously it's in your title and it's something you know a lot about and I'm going to make a statement. You tell me if it's on point or off point. Seems that most of the AI going on in the enterprise is modeling data science applied to troves of data. But there's also a lot of AI going on in consumer, whether it's fingerprint technology or facial recognition or natural language processing. Will, to two-part question, will the consumer market, let's say as it has so often in the enterprise sort of inform us is sort of first part. And then will there be a shift from sort of modeling, if you will, to more, you mentioned autonomous vehicles more AI inferencing in real-time, especially with the Edge. I think you can help us understand that better. >> Yeah, this is a great question. There are three stages to just simplify, I mean, you know, it's probably more sophisticated than that, but let's just simplify there're three stages to building an AI system that ultimately can predict, make a prediction. Or to assist you in decision-making, have an outcome. So you start with the data, massive amounts of data that you have to decide what to feed the machine with. So you feed the machine with this massive chunk of data. And the machine starts to evolve a model based on all the data is seeing it starts to evolve. To a point that using a test set of data that you have separately kept a site that you know the answer for. Then you test the model, you know after you're trained it with all that data to see whether his prediction accuracy is high enough. And once you are satisfied with it, you then deploy the model to make the decision and that's the inference. So a lot of times depending on what we are focusing on. We in data science are we working hard on assembling the right data to feed the machine with? That's the data preparation organization work. And then after which you build your models you have to pick the right models for the decisions and prediction you wanted to make. You pick the right models and then you start feeding the data with it. Sometimes you pick one model and a prediction isn't that a robust, it is good, but then it is not consistent. Now what you do is you try another model. So sometimes you just keep trying different models until you get the right kind, yeah, that gives you a good robust decision-making and prediction. Now, after which, if it's tested well, Q8 you will then take that model and deploy it at the Edge, yeah. And then at the Edge is essentially just looking at new data applying it to the model that you have trained and then that model will give you a prediction or a decision. So it is these three stages, yeah. But more and more, your question reminds me that more and more people are thinking as the Edge become more and more powerful, can you also do learning at the Edge? That's the reason why we spoke about swarm learning the last time, learning at the Edge as a swarm. Because maybe individually they may not have enough power to do so, but as a swarm, they may. >> Is that learning from the Edge or learning at the Edge. In other words, is it-- >> Yes. >> Yeah, you don't understand my question, yeah. >> That's a great question. That's a great question. So answer is learning at the Edge, and also from the Edge, but the main goal, the goal is to learn at the Edge so that you don't have to move the data that Edge sees first back to the Cloud or the call to do the learning. Because that would be the reason, one of the main reasons why you want to learn at the Edge. So that you don't need to have to send all that data back and assemble it back from all the different Edge devices assemble it back to the Cloud side to do the learning. With swarm learning, you can learn it and keep the data at the Edge and learn at that point, yeah. >> And then maybe only selectively send the autonomous vehicle example you gave is great 'cause maybe they're, you know, there may be only persisting. They're not persisting data that is an inclement weather, or when a deer runs across the front and then maybe they do that and then they send that smaller data set back and maybe that's where it's modeling done but the rest can be done at the Edge. It's a new world that's coming to, let me ask you a question. Is there a limit to what data should be collected and how it should be collected? >> That's a great question again, yeah, well, today full of these insightful questions that actually touches on the second challenge. How do we, to in order to thrive in this new age of insight. The second challenge is our future challenge. What do we do for our future? And in there is the statement we make is we have to focus on collecting data strategically for the future of our enterprise. And within that, I talk about what to collect, and when to organize it when you collect, and then where will your data be going forward that you are collecting from? So what, when, and where. For the what data, for what data to collect that was the question you asked. It's a question that different industries have to ask themselves because it will vary. Let me give you the, you use the autonomous car example. Let me use that and you have this customer collecting massive amounts of data. You know, we talking about 10 petabytes a day from a fleet of their cars and these are not production autonomous cars. These are training autonomous cars, collecting data so they can train and eventually deploy a commercial cars. Also these data collection cars, they collect 10 as a fleet of them collect 10 petabytes a day. And then when it came to us, building a storage system to store all of that data they realize they don't want to afford to store all of it. Now here comes the dilemma. What should I, after I spent so much effort building all this cars and sensors and collecting data, I've now decide what to delete. That's a dilemma. Now in working with them on this process of trimming down what they collected. I'm constantly reminded of the 60s and 70s. To remind myself 60s and 70s, we call a large part of our DNA, junk DNA. Today we realized that a large part of that, what we call junk has function has valuable function. They are not genes but they regulate the function of genes. So what's junk in yesterday could be valuable today, or what's junk today could be valuable tomorrow. So there's this tension going on between you deciding not wanting to afford to store everything that you can get your hands on. But on the other hand, you know you worry, you ignore the wrong ones. You can see this tension in our customers. And then it depends on industry here. In healthcare they say, I have no choice. I want it all, why? One very insightful point brought up by one healthcare provider that really touched me was you know, we are not, we don't only care. Of course we care a lot. We care a lot about the people we are caring for. But we also care for the people we are not caring for. How do we find them? And therefore, they did not just need to collect data that they have with, from their patients they also need to reach out to outside data so that they can figure out who they are not caring for. So they want it all. So I asked them, "So what do you do with funding if you want it all?" They say they have no choice but they'll figure out a way to fund it and perhaps monetization of what they have now is the way to come around and fund that. Of course, they also come back to us, rightfully that you know, we have to then work out a way to to help them build a system. So that healthcare. And if you go to other industries like banking, they say they can afford to keep them all. But they are regulated same like healthcare. They are regulated as to privacy and such like. So many examples, different industries having different needs but different approaches to how, what they collect. But there is this constant tension between you perhaps deciding not wanting to fund all of that, all that you can store. But on the other hand you know, if you kind of don't want to afford it and decide not to store some, maybe those some become highly valuable in the future. You worry. >> Well, we can make some assumptions about the future, can't we? I mean we know there's going to be a lot more data than we've ever seen before, we know that. We know, well not withstanding supply constraints and things like NAND. We know the price of storage is going to continue to decline. We also know and not a lot of people are really talking about this but the processing power, everybody says, Moore's Law is dead. Okay, it's waning but the processing power when you combine the CPUs and NPUs, and GPUs and accelerators and so forth, actually is increasing. And so when you think about these use cases at the Edge you're going to have much more processing power. You're going to have cheaper storage and it's going to be less expensive processing. And so as an AI practitioner, what can you do with that? >> Yeah, it's a highly, again another insightful question that we touched on, on our keynote and that goes up to the why, I'll do the where. Where will your data be? We have one estimate that says that by next year, there will be 55 billion connected devices out there. 55 billion. What's the population of the world? Well, off the order of 10 billion, but this thing is 55 billion. And many of them, most of them can collect data. So what do you do? So the amount of data that's going to come in is going to way exceed our drop in storage costs our increasing compute power. So what's the answer? The answer must be knowing that we don't and even a drop in price and increase in bandwidth, it will overwhelm the 5G, it'll will overwhelm 5G, given the amount of 55 billion of them collecting. So the answer must be that there needs to be a balance between you needing to bring all that data from the 55 billion devices of the data back out to a central, as a bunch of central cost because you may not be able to afford to do that. Firstly bandwidth, even with 5G and as the, when you still be too expensive given the number of devices out there. You know given storage costs dropping it'll still be too expensive to try and install them all. So the answer must be to start at least to mitigate the problem to some leave most a lot of the data out there. And only send back the pertinent ones, as you said before. But then if you did that then, how are we going to do machine learning at the core and the Cloud side, if you don't have all the data you want rich data to train with. Sometimes you want to a mix of the positive type data, and the negative type data. So you can train the machine in a more balanced way. So the answer must be you eventually, as we move forward with these huge number of devices are at the Edge to do machine learning at the Edge. Today we don't even have power. The Edge typically is characterized by a lower energy capability and therefore, lower compute power. But soon, you know, even with low energy, they can do more with compute power, improving in energy efficiency. So learning at the Edge today we do inference at the Edge. So we data, model, deploy and you do inference at age. That's what we do today. But more and more, I believe given a massive amount of data at the Edge you have to have to start doing machine learning at the Edge. And if when you don't have enough power then you aggregate multiple devices' compute power into a swarm and learn as a swarm. >> Oh, interesting, so now of course, if I were sitting in a flyer flying the wall on HPE Board meeting I said, "Okay, HPE is a leading provider of compute." How do you take advantage that? I mean, we're going, I know it's future but you must be thinking about that and participating in those markets. I know today you are, you have, you know, Edge line and other products, but there's, it seems to me that it's not the general purpose that we've known in the past. It's a new type of specialized computing. How are you thinking about participating in that opportunity for your customers? >> The wall will have to have a balance. Where today the default, well, the more common mode is to collect the data from the Edge and train at some centralized location or number of centralized location. Going forward, given the proliferation of the Edge devices, we'll need a balance, we need both. We need capability at the Cloud side. And it has to be hybrid. And then we need capability on the Edge side. Yeah that we need to build systems that on one hand is Edge-adapted. Meaning they environmentally-adapted because the Edge differently are on it. A lot of times on the outside, they need to be packaging-adapted and also power-adapted. Because typically many of these devices are battery-powered. So you have to build systems that adapts to it. But at the same time, they must not be custom. That's my belief. They must be using standard processes and standard operating system so that they can run a rich set of applications. So yes, that's also the insightful for that. Antonio announced in 2018 for the next four years from 2018, $4 billion invested to strengthen our Edge portfolio our Edge product lines, Edge solutions. >> Dr. Goh, I could go on for hours with you. You're just such a great guest. Let's close. What are you most excited about in the future of certainly HPE, but the industry in general? >> Yeah, I think the excitement is the customers. The diversity of customers and the diversity in the way they have approached their different problems with data strategy. So the excitement is around data strategy. Just like, you know, the statement made for us was so, was profound. And Antonio said we are in the age of insight powered by data. That's the first line. The line that comes after that is as such we are becoming more and more data-centric with data the currency. Now the next step is even more profound. That is, you know, we are going as far as saying that data should not be treated as cost anymore, no. But instead, as an investment in a new asset class called data with value on our balance sheet. This is a step change in thinking that is going to change the way we look at data, the way we value it. So that's a statement. So this is the exciting thing, because for me a CTO of AI, a machine is only as intelligent as the data you feed it with. Data is a source of the machine learning to be intelligent. So that's why when the people start to value data and say that it is an investment when we collect it it is very positive for AI because an AI system gets intelligent, get more intelligence because it has huge amounts of data and a diversity of data. So it'd be great if the community values data. >> Well, are you certainly see it in the valuations of many companies these days? And I think increasingly you see it on the income statement, you know data products and people monetizing data services, and yeah, maybe eventually you'll see it in the balance sheet, I know. Doug Laney when he was at Gartner Group wrote a book about this and a lot of people are thinking about it. That's a big change, isn't it? Dr. Goh. >> Yeah, yeah, yeah. Your question is the process and methods in valuation. But I believe we'll get there. We need to get started and then we'll get there, I believe, yeah. >> Dr. Goh it's always my pleasure. >> And then the AI will benefit greatly from it. >> Oh yeah, no doubt. People will better understand how to align some of these technology investments. Dr. Goh, great to see you again. Thanks so much for coming back in theCube. It's been a real pleasure. >> Yes, a system is only as smart as the data you feed it with. (both chuckling) >> Well, excellent, we'll leave it there. Thank you for spending some time with us so keep it right there for more great interviews from HPE Discover '21. This is Dave Vellante for theCube, the leader in enterprise tech coverage. We'll be right back (upbeat music)

(upbeat music) >> Hi everybody, this is Dave Volante. And with me is Sandeep Singh, he is the vice president of Storage Marketing at Hewlett Packard Enterprise. And we're going to riff on some of the trends in the industry, what we're seeing. And we got a little treat for you. Sandeep, great to see you man. >> Dave, it's a pleasure to be here. >> You and I've known each other for a long time. We've had some great discussions, some debates, some intriguing mind benders. What are you seeing out there in Storage? So much has changed. What are the key trends you're seeing and let's get into it. >> Yeah, across the board, as you said, so much has changed. When you reflect back at the underlying transformation that's taken place with data, cloud and AI across the board. First of all, for our customers they're seeing this massive data explosion that literally now spans edge to core to cloud. They're also seeing a diversity of the application workloads across the board. And the emphasis that it's placing is on the complexity that underlies overall infrastructure and data management. Across the board, we're hearing a lot from customers about just the underlying infrastructure complexity and the infrastructure sprawl. And then the second element of that is really extending into the complexity of data management. >> So it's interesting you're talking about data management. You remember you and I, we were in Andover. It was probably like five years ago and all we were talking about was media. Flash this and flash that, and at the time that was kind of the hot storage topic. Well, flash came in addressing some of the mics that we historically talked about it. Now the problem statement is really kind of quote unquote metaphorically moving up the stack if you will, you mentioned management but let's dig into that a little bit. I mean, what is management? I mean, a lot of people that means different things to different people. You talk to a database person or a backup person. How do you look at management? What does that mean to you? >> Yeah, Dave, you mentioned that the flash came in and it actually accelerated the overall speed and latency that storage was delivering to the application workloads. But fundamentally when you look back at storage over a couple of decades the underlying way of how you're managing storage hasn't fundamentally changed. There's still an incredible amount of complexity for IT. It's still a manual admin driven experience for customers. And what that's translating to is more often than not IT is in the world of firefighting and it's leaves them unable to help with them more strategic projects to innovate for the business. And basically IT has that pressure point of moving beyond that and helping bring greater levels of agility that line of business owners are asking for and to be able to deliver on more of the strategic projects. So that's one element of it. The second element that we're hearing from customers about is as more and more data just continues to explode from edge to core to cloud. And as basically the infrastructure has grown from just being on-Prem to being at the Edge to being in the cloud. Now that complexity is expanding from just being on-Prem to across multiple different clouds. So when you look across the date data life cycle how do you store it? How do you secure it? How do you basically protect it and archive it and analyze that data. That end to end life cycle management of data today resides on just a fragmented set of overall infrastructure and tools and processes and administrative boundaries. That's creating a massive challenge for customers. And the impact of that ultimately is essentially comes at a cost to agility, to innovation and ultimately business risk. >> Yeah, so we've seen obviously the cloud has addressed a lot of these problems but the problem is the cloud is in the cloud and much of my stuff, most of my stuff, isn't in the cloud. So I have all these other workloads that are either on-Prem and now you've got this emerging Edge. And so I wonder if we could just talk a little vision here for a minute. I mean what I've been envisioning is this abstraction layer that cuts across all weather. It doesn't really matter where it is. If it's on-Prem, if it's across cloud, if it's in the cloud, on the edge, we could talk about what that all means. But if customers that I talked to they're sort of done with the complexity of that underlying infrastructure. They want technology to take care of that. They want automation they want AI brought in to that equation. And it seems like we're from the cusp of the decade where that might happen. What's your take? >> Well, yeah, certainly I mentioned that data cloud and AI are really the disruptive forces, better propelling. The digital transformation for customers. Cloud has set the standard for agility and AI driven insights and intelligence are really helping to make the underlying infrastructure invisible and customers are looking for this notion of being able to get that cloud operational agility pretty much everywhere because they're discovering that that's a game changer. And yet a lot of their application workloads and data is on-Prem and is increasingly growing at the edge. So they want same experience to be able to truly bring that agility to wherever their data in absolute. And that's one of the things that we're continuing to hear from customers. >> And this problem is just going to get worse. I mean for decades we marched to the cadence of Moore's Law and everybody's going to forgets about Moore's Law. And say, "Ah, it's dying or whatever." But actually when you look at the processing power that's coming out now, it's more than doubling every two years, quadrupling every two years. So now you've got this capability in your hands and application design minors, storage companies, networking companies. They're going to have all this power to now bring in AI and do things that we've never even imagined before. So it's not about the box and the speeds and feeds of the box. It's really more about this abstraction layer that I was talking about. The management if you will that you were discussing and what we can do in terms of being able to power new workloads in machine intelligence, it's this kind of ubiquitous, call it the cloud but it's expanding pretty much everywhere in every part of our lives even to the edge you think about autonomous vehicles, you think about factories it's actually quite mind boggling where we're headed. >> It is and you touched upon AI. And certainly when you look at infrastructure, for example there's been a ton of complexity in infrastructure management. One of the studies that was done actually by IDC indicated that over 90% of the challenges that arise, for example ultimately down at the storage infrastructure layer that's powering the apps ultimately arises from way above the stack all the way from the server layer on down where even the virtual machine layer. And there, for example, AIOps for infrastructure has become a game changer for customers to be able to bring the power of AI and machine learning and multi-variate analysis to be able to predict and prevent issues. Dave, you also touched upon Edge and across the board. What we're seeing is the Enterprise Edge is becoming that frontier for customer experiences and the opportunity to reimagine customer experiences as well as just the frontier for commerce that's happening. When you look at retail and manufacturing and or financial services. So across the board with the data growth that's happening and this Edge becoming the strategic frontier for delivering the customer experiences how you power your application workloads there and how you deliver that data and protect that data and be able to seamlessly manage that overall infrastructure. As you mentioned abstracted away at a higher level becomes incredibly important for customers. >> So interesting to hear how the conversations changed. I'd like to say, I go back to whatever it was five years ago, we're talking about flash storage class memory, NVMe and those things are still there but your emphasis now you're talking about machine learning, AI, math around deep learning. It's really software is really what you're focusing on these days. >> Very much so. Certainly this notion of software and services that are delivering and unlocking a whole new experience for customers that's really the game changer going forward for customers. And that's what we're focused on. >> Well, I said we had a little surprise for you. So you guys are having an event on May 4th. It's called Unleash The Power of Data. What's that event all about Sandeep? >> Yeah. We are very much excited about our May 4th event. As you mentioned, it's called Unleash The Power of Data. And as most organizations today are data driven and data is at the heart of what they're doing. We're excited to invite everyone to join this event. And through this event we're unveiling a new vision for data that accelerates the data driven transformation from Edge to cloud. This event promises to be a pivotal event and one that IT admins, cloud architects, virtual machine admins, vice presidents, directors of IT and CIO really won't want to mess. Across the board this event is just bringing a new way of articulating the overall problem statement and in market in focused the articulation of the trends that we were just discussing. It's an event that's going to be hosted by a Business and Technology Journalist, Shabani Joshi. It will feature a market in panel with a focus on the crucial role that data is playing in customers digital transformation. It will also include and feature Antonio Neary, CEO of HPE and Tom black, senior vice president and general manager of HPE Storage Business and industry experts including Julia Palmer, research vice president at Gartner. We will unveil game changing HPE innovations that will make it possible for organizations across Edge to cloud to unleash the power of data. >> Sounds like great event. I presume I can go to hpe.com and get information, is it a registered event? How does that all work? Yeah, we invite everyone to visit hpe.com and by visiting there you can click and save the date of May 4th at 8:00 AM Pacific. We invite everyone to join us. We couldn't be more excited to get to this event and be able to share the vision and game-changing HPE innovations. >> Awesome. So I don't have to register, right? I don't have to give up my three children's name and my social security number to attend your event. Is that right? >> No registration required, come by, click on hpe.com. Save the date on your calendar. And we very much look forward to having everyone join us for this event. >> I love it, it's pure content event. I'm not going to get a phone call afterwards saying, "Hey, buy some stuff from me." That could come other channels but so that's good. Thank you for that. Thanks for providing that service to the industry. I'm excited to see what you guys are going to be announcing that day and look Sandeep. I mean, like I said, we've known each other a while. We've seen a lot of trends but the next 10 years it ain't going to look like the last 10 is it? >> It's going to be very different and we couldn't be more excited. >> Well, Sandeep, thanks so much for coming to theCube and riffing with me on the industry and giving us a preview for your event. Good luck with that. And always great to see you. >> Thanks a lot, Dave. Always great to see you as well. >> All right. And thank you everybody. This is Dave Volante for theCube and we'll see you next time. (upbeat music)

>>from around the globe. It's the Cube presenting Cuban cloud brought to you by silicon angle. As I've said many times on the Cube for years, decades, even we've marched to the cadence of Moore's law, relying on the doubling of performance every 18 months or so. But no longer is this the mainspring of innovation for technology. Rather, it's the combination of data applying machine intelligence and the cloud supported by the relentless reduction of the cost of compute and storage and the build out of a massively distributed computer network. Very importantly, in the last several years, alternative processors have emerged to support offloading work and performing specific Test GP use of the most widely known example of this trend, with the ascendancy of in video for certain applications like gaming and crypto mining and, more recently, machine learning. But in the middle of last decade, we saw the early development focused on the DPU, the data processing unit, which is projected to make a huge impact on data centers in the coming years. As we move into the next era of Cloud. And with me is deep. Sindhu, who's this co founder and CEO of Fungible, a company specializing in the design and development of GPU deep Welcome to the Cube. Great to see you. >>Thank you, Dave. And thank you for having me. >>You're very welcome. So okay, my first question is, don't CPUs and GP use process data already? Why do we need a DPU? >>Um you know that that is a natural question to ask on. CPUs have been around in one form or another for almost, you know, 55 maybe 60 years. And, uh, you know, this is when general purpose computing was invented, and essentially all CPI use went to x 80 60 x 86 architecture. Uh, by and large arm, of course, is used very heavily in mobile computing, but x 86 primarily used in data center, which is our focus. Um, now, you can understand that that architectural off general purpose CPUs has been refined heavily by some of the smartest people on the planet. And for the longest time, uh, improvements you refer the Moore's Law, which is really the improvements off the price performance off silicon over time. Um, that, combined with architectural improvements, was the thing that was pushing us forward. Well, what has happened is that the architectural refinements are more or less done. Uh, you're not going to get very much. You're not going to squeeze more blood out of that storm from the general purpose computer architectures. What has also happened over the last decade is that Moore's law, which is essentially the doubling off the number of transistors, um, on a chip has slowed down considerably on and to the point where you're only getting maybe 10 20% improvements every generation in speed off the grandest er. If that. And what's happening also is that the spacing between successive generations of technology is actually increasing from 2, 2.5 years to now three, maybe even four years. And this is because we are reaching some physical limits in Seamus. Thes limits are well recognized, and we have to understand that these limits apply not just to general purpose if use, but they also apply to GP use now. General purpose, if used, do one kind of confrontation. They really general on bacon do lots and lots of different things. It is actually a very, very powerful engine, Um, and then the problem is it's not powerful enough to handle all computations. So this is why you ended up having a different kind of processor called the GPU, which specializes in executing vector floating point arithmetic operations much, much better than CPL. Maybe 2030 40 times better. Well, GPS have now been around for probably 15, 20 years, mostly addressing graphics computations. But recently, in the last decade or so, they have been used heavily for AI and analytics computations. So now the question is, why do you need another specialized engine called the DPU? Well, I started down this journey about almost eight years ago, and I recognize I was still at Juniper Networks, which is another company that I found it. I recognize that in the data center, um, as the workload changes due to addressing Mawr and Mawr, larger and larger corpus is of data number one. And as people use scale out as the standard technique for building applications, what happens is that the amount of East West traffic increases greatly. And what happens is that you now have a new type off workload which is coming, and today probably 30% off the workload in a data center is what we call data centric. I want to give you some examples of what is the data centric E? >>Well, I wonder if I could interrupt you for a second, because Because I want you to. I want those examples, and I want you to tie it into the cloud because that's kind of the topic that we're talking about today and how you see that evolving. It's a key question that we're trying to answer in this program. Of course, Early Cloud was about infrastructure, a little compute storage, networking. And now we have to get to your point all this data in the cloud and we're seeing, by the way, the definition of cloud expand into this distributed or I think the term you use is disaggregated network of computers. So you're a technology visionary, And I wonder, you know how you see that evolving and then please work in your examples of that critical workload that data centric workload >>absolutely happy to do that. So, you know, if you look at the architectural off cloud data centers, um, the single most important invention was scale out scale out off identical or near identical servers, all connected to a standard i p Internet network. That's that's the architectural. Now, the building blocks of this architecture er is, uh, Internet switches, which make up the network i p Internet switches. And then the servers all built using general purpose X 86 CPUs with D ram with SSD with hard drives all connected, uh, inside the CPU. Now, the fact that you scale these, uh, server nodes as they're called out, um, was very, very important in addressing the problem of how do you build very large scale infrastructure using general purpose computer? But this architectures, Dave, is it compute centric architectures and the reason it's a compute centric architectures. If you open this a server node, what you see is a connection to the network, typically with a simple network interface card. And then you have CP use, which are in the middle of the action. Not only are the CPUs processing the application workload, but they're processing all of the aisle workload, what we call data centric workload. And so when you connect SSD and hard drives and GPU that everything to the CPU, um, as well as to the network, you can now imagine that the CPUs is doing to functions it z running the applications, but it's also playing traffic cop for the I O. So every Io has to go to the CPU and you're executing instructions typically in the operating system, and you're interrupting the CPU many, many millions of times a second now. General Purpose CPUs and the architecture of the CPS was never designed to play traffic cop, because the traffic cop function is a function that requires you to be interrupted very, very frequently. So it's. It's critical that in this new architecture, where there's a lot of data, a lot of East West traffic, the percentage of work clothes, which is data centric, has gone from maybe 1 to 2% to 30 to 40%. I'll give you some numbers, which are absolutely stunning if you go back to, say, 1987 and which is, which is the year in which I bought my first personal computer. Um, the network was some 30 times slower. Then the CPI. The CPI was running at 50 megahertz. The network was running at three megabits per second. Well, today the network runs at 100 gigabits per second and the CPU clock speed off. A single core is about 3 to 2.3 gigahertz. So you've seen that there is a 600 x change in the ratio off I'll to compute just the raw clock speed. Now you can tell me that. Hey, um, typical CPUs have lots of lots, of course, but even when you factor that in, there's bean close toe two orders of magnitude change in the amount of ill to compute. There is no way toe address that without changing the architectures on this is where the DPU comes in on the DPU actually solves two fundamental problems in cloud data centers on these air. Fundamental. There's no escaping it, no amount off. Clever marketing is going to get around these problems. Problem number one is that in a compute centric cloud architectures the interactions between server notes are very inefficient. Okay, that's number one problem number one. Problem number two is that these data center computations and I'll give you those four examples the network stack, the storage stack, the virtualization stack and the security stack. Those four examples are executed very inefficiently by CBS. Needless to say that that if you try to execute these on GPS, you'll run into the same problem, probably even worse because GPS are not good at executing these data centric computations. So when U. S o What we were looking to do it fungible is to solve these two basic problems and you don't solve them by by just using taking older architectures off the shelf and applying them to these problems because this is what people have been doing for the for the last 40 years. So what we did was we created this new microprocessor that we call the DPU from ground doctor is a clean sheet design and it solve those two problems. Fundamental. >>So I want to get into that. But I just want to stop you for a second and just ask you a basic question, which is so if I understand it correctly, if I just took the traditional scale out, If I scale out compute and storage, you're saying I'm gonna hit a diminishing returns, It z Not only is it not going to scale linear linearly, I'm gonna get inefficiencies. And that's really the problem that you're solving. Is that correct? >>That is correct. And you know this problem uh, the workloads that we have today are very data heavy. You take a I, for example, you take analytics, for example. It's well known that for a I training, the larger the corpus of data relevant data that you're training on, the better the result. So you can imagine where this is going to go, especially when people have figured out a formula that, hey, the more data I collect, I can use those insights to make money. >>Yeah, this is why this is why I wanted to talk to you, because the last 10 years we've been collecting all this data. Now I want to bring in some other data that you actually shared with me beforehand. Some market trends that you guys cited in your research and the first thing people said is they want to improve their infrastructure on. They want to do that by moving to the cloud, and they also there was a security angle there as well. That's a whole nother topic. We could discuss the other staff that jumped out at me. There's 80% of the customers that you surveyed said they'll be augmenting their X 86 CPUs with alternative processing technology. So that's sort of, you know, I know it's self serving, but z right on the conversation we're having. So I >>want to >>understand the architecture. Er, aan den, how you've approached this, You've you've said you've clearly laid out the X 86 is not going to solve this problem. And even GP use are not going to solve this problem. So help us understand the architecture and how you do solve this problem. >>I'll be I'll be very happy to remember I use this term traffic cough. Andi, I use this term very specifically because, uh, first let me define what I mean by a data centric computation because that's the essence off the problem resolved. Remember, I said two problems. One is we execute data centric work clothes, at least in order of magnitude, more efficiently than CPUs or GPS, probably 30 times more efficiently on. The second thing is that we allow notes to interact with each other over the network much, much more efficiently. Okay, so let's keep those two things in mind. So first, let's look at the data centric piece, the data centric piece, um, for for workload to qualify as being data centric. Four things have to be true. First of all, it needs to come over the network in the form of packets. Well, this is all workloads, so I'm not saying anything new. Secondly, uh, this workload is heavily multiplex in that there are many, many, many computations that are happening concurrently. Thousands of them. Yeah, that's number two. So a lot of multiplexing number three is that this workload is state fel. In other words, you have to you can't process back. It's out of order. You have to do them in order because you're terminating network sessions on the last one Is that when you look at the actual computation, the ratio off I Oto arithmetic is medium to high. When you put all four of them together, you actually have a data centric workout, right? And this workload is terrible for general purpose, C p s not only the general purpose, C p is not executed properly. The application that is running on the CPU also suffers because data center workloads are interfering workloads. So unless you designed specifically to them, you're going to be in trouble. So what did we do? Well, what we did was our architecture consists off very, very heavily multi threaded, general purpose CPUs combined with very heavily threaded specific accelerators. I'll give you examples of some some of those accelerators, um, de Emma accelerators, then radio coding accelerators, compression accelerators, crypto accelerators, um, compression accelerators, thes air, just something. And then look up accelerators. These air functions that if you do not specialized, you're not going to execute them efficiently. But you cannot just put accelerators in there. These accelerators have to be multi threaded to handle. You know, we have something like 1000 different threads inside our DPU toe address. These many, many, many computations that are happening concurrently but handle them efficiently. Now, the thing that that is very important to understand is that given the paucity off transistors, I know that we have hundreds of billions of transistors on a chip. But the problem is that those transistors are used very inefficiently today. If the architecture, the architecture of the CPU or GPU, what we have done is we've improved the efficiency of those transistors by 30 times. Yeah, so you can use >>the real estate. You can use their real estate more effectively, >>much more effectively because we were not trying to solve a general purpose computing problem. Because if you do that, you know, we're gonna end up in the same bucket where General Focus CPS are today. We were trying to solve the specific problem off data centric computations on off improving the note to note efficiency. So let me go to Point number two, because that's equally important, because in a scale out architecture, the whole idea is that I have many, many notes and they're connected over a high performance network. It might be shocking for your listeners to hear that these networks today run at a utilization of no more than 20 to 25%. Question is why? Well, the reason is that if I tried to run them faster than that, you start to get back. It drops because there are some fundamental problems caused by congestion on the network, which are unsolved as we speak today. There only one solution, which is to use DCP well. DCP is a well known is part of the D. C. P I. P. Suite. DCP was never designed to handle the agencies and speeds inside data center. It's a wonderful protocol, but it was invented 42 year 43 years ago, now >>very reliable and tested and proven. It's got a good track record, but you're a >>very good track record, unfortunately, eats a lot off CPU cycles. So if you take the idea behind TCP and you say, Okay, what's the essence of TCP? How would you apply to the data center? That's what we've done with what we call F C P, which is a fabric control protocol which we intend toe open way. Intend to publish standards on make it open. And when you do that and you you embed F c p in hardware on top of his standard I P Internet network, you end up with the ability to run at very large scale networks where the utilization of the network is 90 to 95% not 20 to 25% on you end up with solving problems of congestion at the same time. Now, why is this important today that zall geek speak so far? But the reason this stuff is important is that it such a network allows you to disaggregate pool and then virtualized, the most important and expensive resource is in the data center. What are those? It's computer on one side, storage on the other side. And increasingly even things like the Ram wants to be disaggregated in food. Well, if I put everything inside a general purpose server, the problem is that those resource is get stranded because they're they're stuck behind the CPI. Well, once you disaggregate those resources and we're saying hyper disaggregate, the meaning, the hyper and the hyper disaggregate simply means that you can disaggregate almost all the resources >>and then you're gonna re aggregate them, right? I mean, that's >>obviously exactly and the network is the key helping. So the reason the company is called fungible is because we are able to disaggregate virtualized and then pull those resources and you can get, you know, four uh, eso scale out cos you know the large aws Google, etcetera. They have been doing this aggregation and pulling for some time, but because they've been using a compute centric architecture, er that this aggregation is not nearly as efficient as we could make on their off by about a factor of three. When you look at enterprise companies, they're off by any other factor of four. Because the utilization of enterprises typically around 8% off overall infrastructure, the utilization the cloud for A W S and G, C, P and Microsoft is closer to 35 to 40%. So there is a factor off almost, uh, 4 to 8, which you can gain by disaggregated and pulling. >>Okay, so I wanna interrupt again. So thes hyper scaler zehr smart. A lot of engineers and we've seen them. Yeah, you're right. They're using ah, lot of general purpose. But we've seen them, uh, move Make moves toward GP use and and embrace things like arm eso I know, I know you can't name names but you would think that this is with all the data that's in the cloud again Our topic today you would think the hyper scaler zehr all over this >>all the hyper scale is recognized it that the problems that we have articulated are important ones on they're trying to solve them. Uh, with the resource is that they have on all the clever people that they have. So these air recognized problems. However, please note that each of these hyper scale er's has their own legacy now they've been around for 10, 15 years, and so they're not in a position to all of a sudden turn on a dime. This is what happens to all companies at some >>point. Have technical debt. You mean they >>have? I'm not going to say they have technical debt, but they have a certain way of doing things on. They are in love with the compute centric way of doing things. And eventually it will be understood that you need a third element called the DPU to address these problems. Now, of course, you heard the term smart neck, and all your listeners must have heard that term. Well, a smart thing is not a deep you what a smart Nick is. It's simply taking general purpose arm cores put in the network interface on a PC interface and integrating them all in the same chip and separating them from the CPI. So this does solve the problem. It solves the problem off the data centric workload, interfering with the application work, work. Good job. But it does not address the architectural problem. How to execute data centric workloads efficiently. >>Yeah, it reminds me. It reminds me of you I I understand what you're saying. I was gonna ask you about smart. Next. It does. It's almost like a bridge or a Band Aid. It's always reminds me of >>funny >>of throwing, you know, a flash storage on Ah, a disc system that was designed for spinning disk gave you something, but it doesn't solve the fundamental problem. I don't know if it's a valid analogy, but we've seen this in computing for a long time. >>Yeah, this analogy is close because, you know. Okay, so let's let's take hyper scaler X. Okay, one name names. Um, you find that, you know, half my CPUs are twiddling their thumbs because they're executing this data centric workload. Well, what are you going to do? All your code is written in, uh, C c plus plus, um, on x 86. Well, the easiest thing to do is to separate the cores that run this workload. Put it on a different Let's say we use arm simply because you know x 86 licenses are not available to people to build their own CPUs. So arm was available, so they put a bunch of encores. Let's stick a PC. I express and network interface on you. Port that quote from X 86 Tow arm. Not difficult to do, but it does yield you results on, By the way, if, for example, um, this hyper scaler X shall we call them if they're able to remove 20% of the workload from general purpose CPUs? That's worth billions of dollars. So of course you're going to do that. It requires relatively little innovation other than toe for quote from one place to another place. >>That's what that's what. But that's what I'm saying. I mean, I would think again. The hyper scale is why Why can't they just, you know, do some work and do some engineering and and then give you a call and say, Okay, we're gonna We're gonna attack these workloads together. You know, that's similar to how they brought in GP use. And you're right. It's it's worth billions of dollars. You could see when when the hyper scale is Microsoft and and Azure, uh, and and AWS both announced, I think they depreciated servers now instead of four years. It's five years, and it dropped, like a billion dollars to their bottom line. But why not just work directly with you guys. I mean, Z the logical play. >>Some of them are working with us. So it's not to say that they're not working with us. So you know, all of the hyper scale is they recognize that the technology that we're building is a fundamental that we have something really special, and moreover, it's fully programmable. So you know, the whole trick is you can actually build a lump of hardware that is fixed function. But the difficulty is that in the place where the DPU would sit, which is on the boundary off a server, and the network is literally on that boundary, that place the functionality needs to be programmable. And so the whole trick is how do you come up with an architectural where the functionality is programmable? But it is also very high speed for this particular set of applications. So the analogy with GPS is nearly perfect because GP use, and particularly in video that's implemented or they invented coulda, which is a programming language for GPS on it made them easy to use mirror fully programmable without compromising performance. Well, this is what we're doing with DP use. We've invented a new architectures. We've made them very easy to program. And they're these workloads or not, Workload. The computation that I talked about, which is security virtualization storage and then network. Those four are quintessential examples off data centric, foreclosed on. They're not going away. In fact, they're becoming more and more and more important over time. >>I'm very excited for you guys, I think, and really appreciate deep we're gonna have you back because I really want to get into some of the secret sauce you talked about these accelerators, Erasure coding, crypto accelerators. I want to understand that. I know there's envy me in here. There's a lot of hardware and software and intellectual property, but we're seeing this notion of programmable infrastructure extending now, uh, into this domain, this build out of this I like this term dis aggregated, massive disaggregated network s so hyper disaggregated. Even better. And I would say this on way. I gotta go. But what got us here the last decade is not the same is what's gonna take us through the next decade. Pretty Thanks. Thanks so much for coming on the cube. It's a great company. >>You have it It's really a pleasure to speak with you and get the message of fungible out there. >>E promise. Well, I promise we'll have you back and keep it right there. Everybody, we got more great content coming your way on the Cube on Cloud, This is David. Won't stay right there.

>> Hi, I'm Stu Miniman, and welcome back to theCUBE on Cloud talking about really important topics as to how developers, were changing how they build their applications, where they live, of course, long discussion we've had for a number of years. You know, how do things change in hybrid environments? We've been talking for years, public cloud and private cloud, and really excited for this session. We're going to talk about how edge environment and AI impact that. So happy to welcome back one of our CUBE alumni, Simon Crosby, is currently the Chief Technology Officer with Swim. He's got plenty of viewpoints on AI, the edge and knows the developer world well. Simon, welcome back. Thanks so much for joining us. >> Thank you, Stu, for having me. >> All right, so let's start for a second. Let's talk about developers. You know, it used to be, you know, for years we talked about, you know, what's the level of abstraction we get. Does it sit, you know, do I put it on bare metal? Do I virtualize it? Do I containerize it? Do I make it serverless? A lot of those things, you know that the app developer doesn't want to even think about but location matters a whole lot when we're talking about things like AI where do I have all my data that I could do my training? Where do I actually have to do the processing? And of course, edge just changes by orders of magnitude. Some of the things like latency and where data lives and everything like that. So with that as a setup, would love to get just your framework as to what you're hearing from developers and what we'll get into some of the solutions that you and your team are helping them to do their jobs. >> Well, you're absolutely right, Stu. The data onslaught is very real. Companies that I deal with are facing more and more real-time data from products from their infrastructure, from their partners whatever it happens to be and they need to make decisions rapidly. And the problem that they're facing is that traditional ways of processing that data are too slow. So perhaps the big data approach, which by now is a bit old, it's a bit long in the tooth, where you store data and then you analyze it later, is problematic. First of all, data streams are boundless. So you don't really know when to analyze, but second you can't store it all. And so the store then analyze approach has to change and Swim is trying to do something about this by adopting a process of analyze on the fly, so as data is generated, as you receive events you don't bother to store them. You analyze them, and then if you have to, you store the data, but you need to analyze as you receive data and react immediately to be able to generate reasonable insights or predictions that can drive commerce and decisions in the real world. >> Yeah absolutely. I remember back in the early days of big data, you know, real time got thrown around a little but it was usually I need to react fast enough to make sure we don't lose the customer, react to something, but it was, we gather all the data and let's move compute to the data. Today as you talk about, you know, real time streams are so important. We've been talking about observability for the last couple of years to just really understand the systems and the outputs more than looking back historically at where things were waiting for alerts. So could you give us some examples if you would, as to you know, those streams, you know, what is so important about being able to interact and leverage that data when you need it? And boy, it's great if we can use it then and not have to store it and think about it later, obviously there's some benefits there, because-- >> Well every product nowadays has a CPU, right? And so there's more and more data. And just let me give you an example, Swim processes real-time data from more than a hundred million mobile devices in real time, for a mobile operator. And what we're doing there is we're optimizing connection quality between devices and the network. Now that volume of data is more than four petabytes per day, okay. Now there is simply no way you can ever store that and analyze it later. The interesting thing about this is that if you adopt and analyze, and then if you really have to store architecture, you get to take advantage of Moore's Law. So you're running at CPU memory speeds instead of at disk speed. And so that gives you a million fold speed up, and it also means you don't have the latency problem of reaching out to, or about storage, database, or whatever. And so that reduces costs. So we can do it on about 10% of the infrastructure that they previously had for Hadoop style implementation. >> So, maybe it would help if we just explain. When we say edge people think of a lot of different things, is it, you know an IOT device sitting out at the edge? Are we talking about the Telecom edge? We've been watching AWS for years, you know, spider out their services and into various environments. So when you talk about the type of solutions you're doing and what your customers have, is it the Telecom edge? Is it the actual device edge, you know, where does processing happen and where do these you know, services that work on it live? >> So I think the right way to think about edge is where can you reasonably process the data? And it obviously makes sense to process data at the first opportunity you have, but much data is encrypted between the original device, say, and the application. And so edge as a place doesn't make as much sense as edge as an opportunity to decrypt and analyze data in the clear. So edge computing is not so much a place in my view as the first opportunity you have to process data in the clear and to make sense of it. And then edge makes sense, in terms of latency, by locating, compute, as close as possible to the sources of data, to reduce latency and maximize your ability to get insights and return them to users, you know, quickly. So edge for me often is the cloud. >> Excellent, one of the other things I think about back from, you know, the big data days or even earlier, it was that how long it took to get from the raw data to processing that data, to be able to getting some insight, and then being able to take action. It sure sounds like we're trying to collapse that completely, is that, you know, how do we do that? You know, can we actually, you know, build the system so that we can, you know, in that real time, continuous model that you talk about, you know. Take care of it and move on. >> So one of the wonderful things, one of the wonderful things about cloud computing is that two major abstractions have really served us. And those are rest, which is static disk computing, and databases. And rest means any old server can do the job for me and then the database is just an API call away. The problem with that is that it's desperately slow. So when I say desperately slow, I mean, it's probably thrown away the last 10 years of Moore's law. Just think about it this way. Your CPU runs at gigahertz and the network runs at milliseconds. So by definition, every time you reach out to a data store you're going a million times slower than your CPU. That's terrible. It's absolutely tragic, okay. So a model which is much more effective is to have an in-memory computer architecture in which you engage in staple computation. So instead of having to reach out to a database every time to update the database and whatever, you know, store something, and then fetch it again a few moments later when the next event arrives, you keep state in memory and you compute on the fly as data arrives. And that way you get a million times speed up. You also end up with this tremendous cost reduction because you don't end up with as many instances having to compute, by comparison. So let me give you a quick example. If you go to a traffic.swim.ai you can see the real time state of the traffic infrastructure in Palo Alto. And each one of those intersections is predicting its own future. Now, the volume of data from just a few hundred lights in Palo Alto is about four terabytes a day. And sure you can deal with this in AWS Lambda. There are lots and lots of servers up there. But the problem is that the end to end per event latency is about 100 milliseconds. And, you know, if I'm dealing with 30,000 events a second, that's just too much. So solving that problem with a stateless architecture is extraordinarily expensive, more than $5,000 a month. Whereas the staple architecture which you could think of as an evolution of, you know, something reactive or the actor model, gets you, you know something like a 10th of the cost, okay. So cloud is fabulous for things that need to scale wide but a staple model is required for dealing with things which update you rapidly or regularly about their changes in state. >> Yeah, absolutely. You know, I think about if, I mentioned before AI training models, often, if you look at something like autonomous vehicles, the massive amounts of data that it needs to process, you know, has to happen in the public cloud. But then that gets pushed back down to the end device, in this case it's a car, because it needs to be able to react in real time and gets fed at a regular update, the new training algorithms that it has there. What are you seeing-- >> I have strong reason on this training approach and data science in general, and that is that there aren't enough data scientists or, you know, smart people to train these algorithms, deploy them to the edge and so on. And so there is an alternative worldview which is a much simpler one and that is that relatively simple algorithms deployed at scale to staple representatives, let's call them digital twins of things, can deliver enormous improvements in behavior as things learn for themselves. So the way I think the, at least this edge world, gets smarter is that relatively simple models of things will learn for themselves, create their own futures, based on what they can see and then react. And so this idea that we have lots and lots of data scientists dealing with vast amounts of information in the cloud is suitable for certain algorithms but it doesn't work for the vast majority of applications. >> So where are we with the state of what, what do developers need to think about? You mentioned that there's compute in most devices. That's true, but, you know, do they need some special Nvidia chip set out there? Are there certain programming languages that you are seeing more prevalent, interoperability, give us a little bit of, you know, some tips and tricks for those developing. >> Super, so number one, a staple architecture is fundamental and sure React is well known and there are ACA for example, and Spurling. Swim is another so I'm going to use some language and I would encourage you to look at swimos.org to go from play there. A staple architecture, which allows actors, small concurrent objects to stapely evolve their own state based on updates from the real world is fundamental. By the way, in Swim we use data to build these models. So these little agents, for things, we call them web agents because the object ID is a URI, they stapley evolve by processing their own real-world data, stapley representing it, And then they do this wonderful thing which is build a model on the fly. And they build a model by linking to things that they're related to. So a need section would link to all of its sensors but it would also link to all of its neighbors because the neighbors and linking is like a sub in Pub/Sub, and it allows that web agent then to continually analyze, learn, and predict on the fly. And so every one of these concurrent objects is doing this job of analyzing its own raw data and then predicting from that and streaming the result. So in Swim, you get streamed raw data in and what streams out is predictions, predictions about the future state of the infrastructure. And that's a very powerful staple approach which can run all their memory, no storage required. By the way, it's still persistent, so if you lose a node, you can just come back up and carry on but there's no need to store huge amounts of raw data if you don't need it. And let me just be clear. The volumes of raw data from the real world are staggering, right? So four terabytes a day from Palo Alto, but Las Vegas about 60 terabytes a day from the traffic lights. More than 100 million mobile devices is tens of petabytes per day, which is just too much to store. >> Well, Simon, you've mentioned that we have a shortage when it comes to data scientists and the people that can be involved in those things. How about from the developers side, do most enterprises that you're talking to do they have the skillset? Is the ecosystem mature enough for the company to get involved? What do we need to do looking forward to help companies be able to take advantage of this opportunity? >> Yeah, so there is this huge challenge in terms of, I guess, just cloud native skills. And this is exacerbated the more you get added to. I guess what you could think of is traditional kind of companies, all of whom have tons and tons of data sources. So we need to make it easy and Swim tries to do this by effectively using skills that people already have, Java or JavaScript, and giving them easy ways to develop, deploy, and then run applications without thinking about them. So instead of binding developers to notions of place and where databases are and all that sort of stuff if they can write simple object-oriented programs about things like intersections and push buttons, and pedestrian lights, and inroad loops and so on, and simply relate basic objects in the world to each other then we let data build the model by essentially creating these little concurrent objects for each thing, and they will then link to each other and solve the problem. We end up solving a huge problem for developers too, which is that they don't need to acquire complicated cloud-native skillsets to get to work. >> Well absolutely, Simon, it's something we've been trying to do for a long time is to truly simplify things. Want to let you have the final word. If you look out there, the opportunity, the challenge in the space, what final takeaways would you give to our audience? >> So very simple. If you adopt a staple competing architecture, like Swim, you get to go a million times faster. The applications always have an answer. They analyze, learn and predict on the fly and they go a million times faster. They use 10% less, no, sorry, 10% of the infrastructure of a store than analyze approach. And it's the way of the future. >> Simon Crosby, thanks so much for sharing. Great having you on the program. >> Thank you, Stu. >> And thank you for joining I'm Stu Miniman, thank you, as always, for watching theCUBE.

from the cube studios in Palo Alto in Boston connecting with thought leaders all around the world this is a cube conversation hi buddy welcome back to the cube this is Dave Volante and you're watching a special presentation data ops enacted made possible by IBM you know what's what's happening is the innovation engine in the IT economy is really shifted used to be Moore's Law today it's applying machine intelligence and AI to data really scaling that and operationalizing that new knowledge the challenges that is not so easy to operationalize AI and infuse it into the data pipeline but what we're doing in this program is bringing in practitioners who have actually had a great deal of success in doing just that and I'm really excited to have it Kumal a Himalayan Manali is here she's the executive head of data management or personal and business banking at Standard Bank of South Africa the tomb of length thanks so much for coming in the queue thank you for having me Dave you're very welcome and first of all how you holding up with this this bovid situation how are things in Johannesburg um things in Johannesburg are fine we've been on lockdown now I think it's day 33 if I'm not mistaken lost count and but we're really grateful for the swift action of government we we only I mean we have less than 4,000 places in the country and infection rate is is really slow so we've really I think been able to find the curve and we're grateful for being able to be protected in this way so all working from home or learning the new normal and we're all in this together that's great to hear why don't you tell us a little bit about your your role you're a data person we're really going to get into it but here with us you know how you spend your time okay well I head up a date operations function and a data management function which really is the foundational part of the data value chain that then allows other parts of the organization to monetize data and liberate it as as as the use cases apply we monetize it ourselves as well but really we're an enterprise wide organization that ensures that data quality is managed data is governed that we have the effective practices applied to the entire lineage of the data ownership and curation is in place and everything else from a regulatory as well as opportunity perspective then is able to be leveraged upon so historically you know data has been viewed as sort of this expense it's it's big it's growing it needs to be managed deleted after a certain amount of time and then you know ten years ago of the Big Data move data became an asset you had a lot of shadow I people going off and doing things that maybe didn't comply to the corporate ethics probably drove here here you're a part of the organization crazy but talk about that how what has changed but they in the last you know five years or so just in terms of how people approach data oh I mean you know the story I tell my colleague who are all bankers obviously is the fact that the banker in 1989 had to mainly just know debits credits and be able to look someone in the eye and know whether or not they'd be a credit risk or not you know if we lend you money and you pay it back the the banker of the late 90s had to then contend with the emergence of technologies that made their lives easier and allowed for automation and processes to run much more smoothly um in the early two-thousands I would say that digitization was a big focus and in fact my previous role was head of digital banking and at the time we thought digital was the panacea it is the be-all and end-all it's the thing that's gonna make organizations edit lo and behold we realized that once you've gotten all your digital platforms ready they are just the plate or the pipe and nothing is flowing through it and there's no food on the face if data is not the main photo really um it's always been an asset I think organizations just never consciously knew that data was that okay so so what sounds like once you've made that sort of initial digital transformation you really had to work it and what we're hearing from a lot of practitioners like self as challenges related to that involve different parts of the organization different skill sets of challenges and sort of getting everybody to work together on the same page it's better but maybe you could take us back to sort of when you started on this initiative around data Ops what was that like what were some of the challenges that you faced and how'd you get through them okay first and foremost Dave organizations used to believe that data was I t's problem and that's probably why you you then saw the emergence of things like chatter IP but when you really acknowledge that data is an essay just like money is an asset then you you have to then take accountability for it just the same way as you would any other asset in the organization and you will not abdicate its management to a separate function that's not cold to the business and oftentimes IT are seen as a support or an enabling but not quite the main show in most organizations right so what we we then did is first emphasize that data is a business capability the business function it presides in business makes to product management makes to marketing makes to everything else that the business needs for data management also has to be for to every role in every function to different degrees and varying bearing offense and when you take accountability as an owner of a business unit you also take accountability for the data in the systems that support the business unit for us that was the first picture um and convincing my colleagues that data was their problem and not something that we had to worry about they just kind of leave us to to it was was also a journey but that was kind of the first step into it in terms of getting the data operations journey going um you had to first acknowledge please carry on no you just had to first acknowledge that it's something you must take accountability of as a banker not just need to a different part of the organization that's a real cultural mindset you know in the game of rock-paper-scissors you know culture kinda beats everything doesn't it it's almost like a yep a trump card and so so the businesses embrace that but but what did you do to support that is there has to be trust in the data that it has to be a timeliness and so maybe you could take us through how you achieve those objectives and maybe some other objectives that business the man so the one thing I didn't mention Dave is that obviously they didn't embrace it in the beginning it wasn't a it wasn't there oh yeah that make sense they do that type of conversation um what what he had was a few very strategic people with the right mindset that I could partner with that understood the case for data management and while we had that as as an in we developed a framework for a fully matured data operations capability in the organization and what that would look like in a target date scenario and then what you do is you wait for a good crisis so we had a little bit of a challenge in that our local regulator found us a little bit wanting in terms of our date of college and from that perspective it then brought the case for data quality management so now there's a burning platform you have an appetite for people to partner with you and say okay we need this to comply to help us out and when they start seeing their opt-in action do they then buy into into the concept so sometimes you need to just wait for a good Christ and leverage it and only do that which the organization will appreciate at that time you don't have to go Big Bang data quality management was the use case at the time five years ago so we focused all our energy on that and after that it gave us leeway and license really bring to maturity all the other capabilities at the business might not well understand as well so when that crisis hit of thinking about people process in technology you probably had to turn some knobs in each of those areas can you talk about that so from a technology perspective that that's when we partnered with with IBM to implement information analyzer for us in terms of making sure that then we could profile the data effectively what was important for us is to to make strides in terms of showing the organization progress but also being able to give them access to self-service tools that will give them insight into their data from a technology perspective that was kind of I think the the genesis of of us implementing and the IBM suite in earnest from a data management perspective people wise we really then also began a data stewardship journey in which we implemented business unit stewards of data I don't like using the word steward because in my organization it's taken lightly almost like a part-time occupation so we converted them we call them data managers and and the analogy I would give is every department with a P&L any department worth its salt has a FDA or financial director and if money is important to you you have somebody helping you take accountability and execute on your responsibilities in managing that that money so if data is equally important as an asset you will have a leader a manager helping you execute on your data ownership accountability and that was the people journey so firstly I had kind of soldiers planted in each department which were data managers that would then continue building the culture maturing the data practices as as applicable to each business unit use cases so what was important is that every manager in every business unit to the Data Manager focus their energy on making that business unit happy by ensuring that they data was of the right compliance level and the right quality the right best practices from a process and management perspective and was governed and then in terms of process really it's about spreading through the entire ecosystem data management as a practice and can be quite lonely um in the sense that unless the whole business of an organization is managing data they worried about doing what they do to make money and most people in most business units will be the only unicorn relative to everybody else who does what they do and so for us it was important to have a community of practice a process where all the data managers across business as well as the technology parts and the specialists who were data management professionals coming together and making sure that we we work together on on specific you say so I wonder if I can ask you so the the industry sort of likes to market this notion of of DevOps applied to data and data op have you applied that type of mindset approach agile of continuous improvement is I'm trying to understand how much is marketing and how much actually applicable in the real world can you share well you know when I was reflecting on this before this interview I realized that our very first use case of data officers probably when we implemented information analyzer in our business unit simply because it was the first time that IT and business as well as data professionals came together to spec the use case and then we would literally in an agile fashion with a multidisciplinary team come together to make sure that we got the outcomes that we required I mean for you to to firstly get a data quality management paradigm where we moved from 6% quality at some point from our client data now we're sitting at 99 percent and that 1% literally is just the timing issue to get from from 6 to 99 you have to make sure that the entire value chain is engaged so our business partners will the fundamental determinant of the business rules apply in terms of what does quality mean what are the criteria of quality and then what we do is translate that into what we put in the catalog and ensure that the profiling rules that we run are against those business rules that were defined at first so you'd have upfront determination of the outcome with business and then the team would go into an agile cycle of maybe two-week sprints where we develop certain things have stand-ups come together and then the output would be - boarded in a prototype in a fashion where business then gets to go double check that out so that was the first iterate and I would say we've become much more mature at it and we've got many more use cases now and there's actually one that it's quite exciting that we we recently achieved over the end of of 2019 into the beginning of this year so what we did was they I'm worried about the sunlight I mean through the window you look creative to me like sunset in South Africa we've been on the we've been on CubeSat sometimes it's so bright we have to put on sunglasses but so the most recent one which was in in mates 2019 coming in too early this year we we had long kind of achieved the the compliance and regulatory burning platform issues and now we are in a place of I think opportunity and luxury where we can now find use cases that are pertinent to business execution and business productivity um the one that comes to mind is we're a hundred and fifty eight years old as an organization right so so this Bank was born before technology it was also born in the days of light no no no integration because every branch was a standalone entity you'd have these big ledges that transactions were documented in and I think once every six months or so these Ledger's would be taken by horse-drawn carriage to a central place to get go reconcile between branches and paper but the point is if that is your legacy the initial kind of ERP implementations would have been focused on process efficiency based on old ways of accounting for transactions and allocating information so it was not optimized for the 21st century our architecture had has had huge legacy burden on it and so going into a place where you can be agile with data is something that we constantly working toward so we get to a place where we have hundreds of branches across the country and all of them obviously telling to client servicing clients as usual and and not being able for any person needing sales teams or executional teams they were not able in a short space of time to see the impact of the tactic from a database fee from a reporting history and we were in a place where in some cases based on how our Ledger's roll up and the reconciliation between various systems and accounts work it would take you six weeks to verify whether your technique were effective or not because to actually see the revenue hitting our our general ledger and our balance sheet might take that long that is an ineffective way to operate in a such a competitive environment so what you had our frontline sales agents literally manually documenting the sales that they had made but not being able to verify whether that or not is bringing revenue until six weeks later so what we did then is we sat down and defined all the requirements were reporting perspective and the objective was moved from six weeks latency to 24 hours um and even 24 hours is not perfect our ideal would be that bite rows of day you're able to see what you've done for that day but that's the next the next epoch that will go through however um we literally had the frontline teams defining what they'd want to see in a dashboard the business teams defining what the business rules behind the quality and the definitions would be and then we had an entire I'm analytics team and the data management team working around sourcing the data optimising and curating it and making sure that the latency had done that's I think only our latest use case for data art um and now we're in a place where people can look at a dashboard it's a cubed self-service they can learn at any time I see the sales they've made which is very important right now at the time of covert nineteen from a form of productivity and executional competitiveness those are two great use cases of women lying so the first one you know going from data quality 6% the 99% I mean 6% is all you do is spend time arguing about the data bills profanity and then 99% you're there and you said it's just basically a timing issue use latency in the timing and then the second one is is instead of paving the cow path with an outdated you know ledger Barret data process week you've now compressed that down to 24 hours you want to get the end of day so you've built in the agility into your data pipeline I'm going to ask you then so when gdpr hit were you able to very quickly leverage this capability and and apply and then maybe other of compliance edik as well well actually you know what we just now was post TDP our us um and and we got GDP all right about three years ago but literally all we got right was reporting for risk and compliance purposes they use cases that we have now are really around business opportunity lists so the risk so we prioritize compliance report a long time it but we're able to do real-time reporting from a single transaction perspective I'm suspicious transactions etc I'm two hours in Bank and our governor so from that perspective that was what was prioritize in the beginning which was the initial crisis so what you found is an entire engine geared towards making sure that data quality was correct for reporting and regulatory purposes but really that is not the be-all and end-all of it and if that's all we did I believe we really would not have succeeded or could have stayed dead we succeeded because Dana monetization is actually the penis' t the leveraging of data for business opportunity is is actually then what tells you whether you've got the right culture or not you're just doing it to comply then it means the hearts and minds of the rest of the business still aren't in the data game I love this story because it's me it's nirvana for so many years we've been pouring money to mitigate risk and you have no choice do it you know the general council signs off on it the the CFO but grudgingly signs off on it but it's got to be done but for years decades we've been waiting to use these these risk initiatives to actually drive business value you know it kind of happened with enterprise data warehouse but it was too slow it was complicated and it certainly didn't happen with with email archiving that was just sort of a tech balk it sounds like you know we're at that point today and I want to ask you I mean like you know you we talking earlier about you know the crisis gonna perpetuated this this cultural shift and you took advantage of that so we're out who we the the mother nature dealt up a crisis like we've never seen before how do you see your data infrastructure your data pipeline your data ops what kind of opportunities do you see in front of you today as a result of ovid 19 well I mean because of of the quality of kind data that we had now we were able to very quickly respond to to pivot nineteen in in our context where the government put us on lockdown relatively early in in the curve or in the cycle of infection and what it meant is it brought a little bit of a shock to the economy because small businesses all of a sudden didn't have a source of revenue or potentially three to six weeks and based on the data quality work that we did before it was actually relatively easy to be agile enough to do the things that we did so within the first weekend of of lockdown in South Africa we were the first bank to proactively and automatically offer small businesses and student and students with loans on our books a instant three month payment holiday assuming they were in good standing and we did that upfront though it was actually an opt-out process rather than you had to fall in and arrange for that to happen and I don't believe we would have been able to do that if our data quality was not with um we have since made many more initiatives to try and keep the economy going to try and keep our clients in in a state of of liquidity and so you know data quality at that point and that Dharma is critical to knowing who you're talking to who needs what and in which solutions would best be fitted towards various segments I think the second component is um you know working from home now brings an entirely different normal right so so if we had not been able to provide productivity dashboard and and and sales and dashboards to to management and all all the users that require it we would not be able to then validate or say what our productivity levels are now that people are working from home I mean we still have essential services workers that physically go into work but a lot of our relationship bankers are operating from home and that face the baseline and the foundation that we said productivity packing for various methods being able to be reported on in a short space of time has been really beneficial the next opportunity for us is we've been really good at doing this for the normal operational and front line and type of workers but knowledge workers have also know not necessarily been big productivity reporters historically they kind of get an output then the output might be six weeks down the line um but in a place where teams now are not locate co-located and work needs to flow in an edge of passion we need to start using the same foundation and and and data pipeline that we've laid down as a foundation for the reporting of knowledge work and agile team type of metric so in terms of developing new functionality and solutions there's a flow in a multidisciplinary team and how do those solutions get architected in a way where data assists in the flow of information so solutions can be optimally developed well it sounds like you're able to map a metric but business lines care about you know into these dashboards you usually the sort of data mapping approach if you will which makes it much more relevant for the business as you said before they own the data that's got to be a huge business benefit just in terms of again we talked about cultural we talked about speed but but the business impact of being able to do that it has to be pretty substantial it really really is um and and the use cases really are endless because every department finds their own opportunity to utilize in terms of their also I think the accountability factor has has significantly increased because as the owner of a specific domain of data you know that you're not only accountable to yourself and your own operation but people downstream to you as a product and in an outcome depend on you to ensure that the quality of the data you produces is of a high nature so so curation of data is a very important thing and business is really starting to understand that so you know the cards Department knows that they are the owners of card data right and you know the vehicle asset Department knows that they are the owners of vehicle they are linked to a client profile and all of that creates an ecosystem around the plan I mean when you come to a bank you you don't want to be known as a number and you don't want to be known just for one product you want to be known across everything that you do with that with that organization but most banks are not structured that way they still are product houses and product systems on which your data reside and if those don't act in concert then we come across extremely schizophrenic as if we don't know our clients and so that's very very important stupid like I can go on for an hour talking about this topic but unfortunately we're we're out of time thank you so much for sharing your deep knowledge and your story it's really an inspiring one and congratulations on all your success and I guess I'll leave it with you know what's next you gave us you know a glimpse of some of the things you wanted to do pressing some of the the elapsed times and the time cycle but but where do you see this going in the next you know kind of mid term and longer term currently I mean obviously AI is is a big is a big opportunity for all organizations and and you don't get automation of anything right if the foundations are not in place so you believe that this is a great foundation for anything AI to be applied in terms of the use cases that we can find the second one is really providing an API economy where certain data product can be shared with third parties I think that probably where we want to take things as well we are really utilizing external third-party data sources I'm in our data quality management suite to ensure validity of client identity and and and residents and things of that nature but going forward because been picked and banks and other organizations are probably going to partner to to be more competitive going forward we need to be able to provide data product that can then be leveraged by external parties and vice-versa to be like thanks again great having you thank you very much Dave appreciate the opportunity thank you for watching everybody that we go we are digging in the data ops we've got practitioners we've got influencers we've got experts we're going in the crowd chat it's the crowd chat net flash data ops but keep it right there way back but more coverage this is Dave Volante for the cube [Music] you

(upbeat music) >> Narrator: From theCUBE Studios in Palo Alto and Boston, connecting with thought leaders all around the world, this is theCUBE Coversation. >> All right everybody, this is Dave Vellante and welcome to this special CUBE Conversation. I first met Frank Slootman in 2007 when he was the CEO of Data Domain. Back then he was the CEO of a disruptive company and still is. Data Domain, believe or not back then, was actually replacing tape drives as the primary mechanism for backup. Yes, believe it or not, it used to be tape. Fast forward several years later, I met Frank again at VMworld when he had become the CEO of ServiceNow. At the time ServiceNow was a small company, about 100 plus million dollars. Frank and his team took that company to 1.2 billion. And Gartner, at the time of IPO said "you know, this doesn't make sense. "It's a small market, it's a very narrow help desk market, "it's maybe a couple billion dollars." The vision of Slootman and his team was to really expand the total available market and execute like a laser. Which they did and today, ServiceNow a very, very successful company. Snowflake first came into my line of sight in 2015 when SiliconANGLE wrote an article, "Why Snowflake is Better "Than Amazon Redshift, Re-imagining Data". Well last year Frank Slootman joined Snowflake, another disruptive company. And he's here today to talk about how Snowflake is really participating in this COVID-19 crisis. And I really want to share some of Frank's insights and leadership principles, Frank great to see you, thanks for coming on. >> Yeah, thanks for having us Dave. >> So when I first reported earlier this year on Snowflake and shared some data with the community, you reached back out to me and said "Dave, I want to just share with you. "I am not a playbook CEO, I am a situational CEO. "This is what I learned in the military." So Frank, this COVID-19 situation was thrown at you, it's a black swan, what was your first move as a leader? >> Well, my first move is let's not overreact. Take a deep breath. Let's really examine what we know. Let's not jump to conclusions, let's not try to project things that we're not capable of projecting. That's hard because we tend to have sort of levels of certainty about what's going to happen in the week, in the next month and so on and all of a sudden that's out of the window. It creates enormous anxiety with people. So in other words you got to sort of reset to okay, what do we know, what can we do, what do we control? And not let our minds sort of go out of control. So I talk to our people all the time about maintain a sense of normalcy, focus on the work, stay in the moment and by the way, turn the newsfeed off, right, because the hysteria you get fed through the media is really not helpful, right? So just cool down and focus on what we still can do. And then I think then everybody takes a deep breath and we just go back to work. I mean, we're in this mode now for three weeks and I can tell you, I'm on teleconferencing calls, whatever, eight, nine hours a day. Prospects, customers, all over the world. Pretty much what I was doing before except I'm not traveling right now. So it's not, >> Yeah, so it sounds clear-- >> Not that different than what it was before. (laughs) >> It sounds very Bill Belichickian, you know? >> Yeah. >> Focus on those things of which you can control. When you were running ServiceNow I really learned it from you and of course Mike Scarpelli, your then and current CFO about the importance of transparency. And I'm interested in how you're communicating, it sounds like you're doing some very similar things but have you changed the way in which you've communicated to your team, your internal employees at all? >> We're communicating much more. Because we can no longer rely on sort of running into people here, there and everywhere. So we have to be much more purposeful about communications. For example, I mean I send an email out to the entire company on Monday morning. And it's kind of a bunch of anecdotes. Just to bring the connection back, the normalcy. It just helps people get connected back to the mothership and like well, things are still going on. We're still talking in the way we always used to be. And that really helps and I also, I check in with people a lot more, I ask all of our leadership to constantly check in with people because you can't assume that everybody is okay, you can't be out of sight, out of mind. So we need to be more purposeful in reaching out and communicating with people than we were previously. >> And a lot of people obviously concerned about their jobs. Have you sort of communicated, what have you communicated to employees about layoffs? I mean, you guys just did a large raise just before all this, your timing was kind of impeccable. But what have you communicated in that regard? >> I've said, there's no layoffs on our radar, number one. Number two, we are hiring. And number three is we have a higher level of scrutiny on the hires that we're making. And I am very transparent. In other words I tell people look, I prioritize the roles that are closest to the direct train of the business. Right, it's kind of common sense. But I wanted to make sure that this is how we're thinking about it. There are some roles that are more postponable than others. I'm hiring in engineering without any reservation because that is the long term strategic interest of the company. One the sales side, I want to know that sales leaders know how to convert to yields, that we're not just sort of bringing capacity online. And the leadership is not convinced or confident that they can convert to yield. So there's a little bit finer level of scrutiny on the hiring. But by and large, it's not that different. There's this saying out there that we should suspend all non-essential spending and hiring, I'm like you should always do that. Right? I mean what's different today? (both laugh) If it's non-essential, why do it, right? So all of this comes back to this is probably how we should operate anyways, yep. >> I want to talk a little bit about the tech behind Snowflake. I'm very sensitive when CEOs come on my program to make sure that we're not, I'm not trying to bait CEOs into ambulance chasing, that's not what it's about. But I do want to share with our community kind of what's new, what's changed and how companies like Snowflake are participating in this crisis. And in particular, we've been reporting for awhile, if you guys bring up that first slide. That the innovation in the industry is really no longer about Moore's Law. It's really shifted. There's a new, what we call an innovation cocktail in the business and we've collected all this data over the last 10 years. With Hadoop and other distributed data and now we have Edge Data, et cetera, there's this huge trove of data. And now AI is becoming real, it's becoming much more economical. So applying machine intelligence to this data and then the Cloud allows us to do this at scale. It allows us to bring in more data sources. It brings an agility in. So I wonder if you could talk about sort of this premise and how you guys fit. >> Yeah, I would start off by reordering the sequence and saying Cloud's number one. That is foundational. That helps us bring scale to data that we never had to number two, it helps us bring computational power to data at levels we've never had before. And that just means that queries and workloads can complete orders of magnitude faster than they ever could before. And that introduces concepts like the time value of data, right? The faster you get it, the more impactful and powerful it is. I do agree, I view AI as sort of the next generation of analytics. Instead of using data to inform people, we're using data to drive processes and businesses directly, right? So I'm agreeing obviously with these strengths because we're the principal beneficiaries and drivers of these platforms. >> Well when we talked about earlier this year about Snowflake, we really brought up the notion that you guys were one of the first if not the first. And guys, bring back Frank, I got to see him. (Frank chuckles) One of the first to really sort of separate the notion of being able to scale, compute independent of storage. And that brought not only economics but it brought flexibility. So you've got this Cloud-native database. Again, what caught my attention in that Redshift article we wrote is essentially for our audience, Redshift was based on ParAccel. Amazon did a great job of really sort of making that a Cloud database but it really wasn't born in the Cloud and that's sort of the advantage of Snowflake. So that architectural approach is starting to really take hold. So I want to give an example. Guys if you bring up the next chart. This is an example of a system that I've been using since early January when I saw this COVID come out. Somebody texted me this. And it's the Johns Hopkins dataset, it's awesome. It shows you, go around the map, you can follow it, it's pretty close to real time. And it's quite good. But the problem is, all right thank you guys. The problem is that when I started to look at, I wanted to get into sort of a more granular view of the counties. And I couldn't do that. So guys bring up the next slide if you would. So what I did was I searched around and I found a New York Times GitHub data instance. And you can see it in the top left here. And basically it was a CSV. And notice what it says, it says we can't make this file beautiful and searchable because it's essentially too big. And then I ran into what you guys are doing with Star Schema, Star Schema's a data company. And essentially you guys made the notion that look, the Johns Hopkins dataset as great as it is it's not sort of ready for analytics, it's got to be cleaned, et cetera. And so I want you to talk about that a little bit. Guys, if you could bring Frank back. And share with us what you guys have done with Star Schema and how that's helping understand COVID-19 and its progression. >> Yeah, one of the really cool concepts I've felt about Snowflake is what we call the data sharing architecture. And what that really means is that if you and I both have Snowflake accounts, even though we work for different institutions, we can share data optics, tables, schema, whatever they are with each other. And you can process against that in place if they are residing in a local, to your own platform. We have taken that concept from private also to public. So that data providers like Star Schema can list their datasets, because they're a data company, so obviously it's in their business interest to allow this data to be profiled and to be accessible by the Snowflake community. And this data is what we call analytics ready. It is instantly accessible. It is also continually updated, you have to do nothing. It's augmented with incremental data and then our Snowflake users can just combine this data with supply chain, with economic data, with internal operating data and so on. And we got a very strong reaction from our customer base because they're like "man, you're saving us weeks "if not months just getting prepared to start to do an al, let alone doing them." Right? Because the data is analytics ready and they have to do literally nothing. I mean in other words if they ask us for it in the morning, in the afternoon they'll be running workloads again. Right, and then combining it with their own data. >> Yeah, so I should point out that that New York Times GitHub dataset that I showed you, it's a couple of days behind. We're talking here about near realtime, or as close as realtime as you can get, is that right? >> Yep. Yeah, every day it gets updated. >> So the other thing, one of the things we've been reporting, and Frank I wondered if you could comment on this, is this new emerging workloads in the Cloud. We've been reporting on this for a couple of years. The first generation of Cloud was IS, was really about compute, storage, some database infrastructure. But really now what we're seeing is these analytic data stores where the valuable data is sitting and much of it is in the Cloud and bringing machine intelligence and data science capabilities to that, to allow for this realtime or near realtime analysis. And that is a new, emerging workload that is really gaining a lot of steam as these companies try to go to this so-called digital transformation. Your comments on that. >> Yeah, we refer to that as the emergence or the rise of the data Cloud. If you look at the Cloud landscape, we're all very familiar with the infrastructure clouds. AWS and Azure and GCP and so on, it's just massive storage and servers. And obviously there's data locked in to those infrastructure clouds as well. We've been familiar for it for 10, 20 years now with application clouds, notably Salesforce but obviously Workday, ServiceNow, SAP and so on, they also have data in them, right? But now you're seeing that people are unsiloing the data. This is super important. Because as long as the data is locked in these infrastructure clouds, in these application clouds, we can't do the things that we need to do with it, right? We have to unsilo it to allow the scale of querying and execution against that data. And you don't see that any more clear that you do right now during this meltdown that we're experiencing. >> Okay so I learned long ago Frank not to argue with you but I want to push you on something. (Frank laughs) So I'm not trying to be argumentative. But one of those silos is on-prem. I've heard you talk about "look, we're a Cloud company. "We're Cloud first, we're Cloud only. "We're not going to do an on-prem version." But some of that data lives on-prem. There are companies out there that are saying "hey, we separate compute and storage too, "we run in the Cloud. "But we also run on-prem, that's our big differentiator." Your thoughts on that. >> Yeah, we burnt the ship behind us. Okay, we're not doing this endless hedging that people have done for 20 years, sort of keeping a leg in both worlds. Forget it, this will only work in the public Cloud. Because this is how the utility model works, right? I think everybody is coming to this realization, right? I mean excuses are running out at this point. We think that it'll, people will come to the public Cloud a lot sooner than we will ever come to the private Cloud. It's not that we can't run on a private cloud, it just diminishes the potential and the value that we bring. >> So as sort of mentioned in my intro, you have always been at the forefront of disruption. And you think about digital transformation. You know Frank we go to all of these events, it used to be physical and now we're doing theCUBE digital. And so everybody talks about digital transformation. CEOs get up, they talk about how they're helping their customers move to digital. But the reality is is when you actually talk to businesses, there was a lot of complacency. "Hey, this isn't really going to happen in my lifetime" or "we're doing pretty well." Or maybe the CEO might be committed but it doesn't necessarily trickle down to the P&L managers who have an update. One of the things that we've been talking about is COVID-19 is going to accelerate that digital transformation and make it a mandate. You're seeing it obviously in retail play out and a number of other industries, supply chains are, this has wreaked havoc on supply chains. And so there's going to be a rethinking. What are your thoughts on the acceleration of digital transformation? >> Well obviously the crisis that we're experiencing is obviously an enormous catalyst for digital transformation and everything that that entails. And what that means and I think as a industry we're just victims of inertia. Right, I mean haven't understood for 20 years why education, both K through 12 but also higher ed, why they're so brick and mortar bound and the way they're doing things, right? And we could massively scale and drop the cost of education by going digital. Now we're forced into it and everybody's like "wow, "this is not bad." You're right, it isn't, right but we haven't so the economics, the economic imperative hasn't really set in but it is now. So these are all great things. Having said that, there are also limits to digital transformation. And I'm sort of experiencing that right now, being on video calls all day. And oftentimes people I've never met before, right? There's still a barrier there, right? It's not like digital can replace absolutely everything. And that is just not true, right? I mean there's some level of filter that just doesn't happen when you're digital. So there's still a need for people to be in the same place. I don't want to sort of over rotate on this concept, that like okay, from here on out we're all going to be on the wires, that's not the way it will be. >> Yeah, be balanced. So earlier you made a comment, that "we should never "be spending on non-essential items". And so you've seen (Frank laughs) back in 2008 you saw the Rest in Peace good times, you've seen the black swan memos that go out. I assume that, I mean you're a very successful investor as well, you've done a couple of stints in the VC community. What are you seeing in the Valley in regard to investments, will investments continue, will we continue to feed innovation, what's your sense of that? Well this is another wake up call. Because in Silicon Valley there's way too much money. There's certainly a lot of ideas but there's not a lot of people that can execute on it. So what happens is a lot of things get funded and the execution is either no good or it's just not a valid opportunity. And when you go through a downturn like this you're finding out that those businesses are not going to make it. I mean when the tide is running out, only the strongest players are going to survive that. It's almost a natural selection process that happens from time to time. It's not necessarily a bad thing because people get reallocated. I mean Silicon Valley is basically one giant beehive, right? I mean we're constantly repurposing money and people and talent and so on. And that's actually good because if an idea is not worth in investing in, let's not do it. Let's repurpose those resources in places where it has merit, where it has viability. >> Well Frank, I want to thank you for coming on. Look, I mean you don't have to do this. You could've retired long, long ago but having leaders like you in place in these times of crisis, but even when in good times to lead companies, inspire people. And we really appreciate what you do for companies, for your employees, for your customers and certainly for our community, so thanks again, I really appreciate it. >> Happy to do it, thanks Dave. >> All right and thank you for watching everybody, Dave Vellante for theCUBE, we will see you next time. (upbeat music)

live from Stanford University it's the hue covering Stanford women in data science 2020 brought to you by Silicon angle media hi and welcome to the cube I'm your host Sonia - Garrett and we're live at Stanford University covering wigs women in data science conference the fifth annual one and joining us today is Daphne Koller who is the co-founder who sari is the CEO and founder of in seat row that Daphne welcome to the cube nice to be here Sonia thank you for having me so tell us a little bit about in seat row how you how it you got it founded and more about your role so I've been working in the intersection of machine learning and biology and health for quite a while and it was always a bit of a an interesting journey in that the data sets were quite small and limited we're now in a different world where there's tools that are allowing us to create massive biological data sets that I think can help us solve really significant societal problems and one of those problems that I think is really important is drug discovery development where despite many important advancements the costs just keep going up and up and up and the question is can we use machine learning to solve that problem better and you talk about this more in your keynote so give us a few highlights of what you talked about so in the last you can think of drug discovery and development in the last 50 to 70 years as being a bit of a glass half-full glass half-empty the glass half-full is the fact that there's diseases that used to be a death sentence or of the sentence still a life long of pain and suffering that are now addressed by some of the modern-day medicines and I think that's absolutely amazing the other side of it is that the cost of developing new drugs has been growing exponentially in what's come to be known as Arun was law being the inverse of Moore's Law which is the one we're all familiar with because the number of drugs approved per billion u.s. dollars just keeps going down exponentially so the question is can we change that curve and you talked in your keynote about the interdisciplinary cold to tell us more about that I think in order to address some of the critical problems that were facing one needs to really build a culture of people who work together at from different disciplines each bringing their own insights and their own ideas into the mix so and in seat row we actually have a company that's half-life scientists many of whom are producing data for the purpose of driving machine learning models and the other half are machine learning people and data scientists who are working on those but it's not a handoff where one group produces the data and the other one consumes and interpreted but really they start from the very beginning to understand what are the problems that one could solve together how do you design the experiment how do you build the model and how do you derive insights from that that can help us make better medicines for people and I also wanted to ask you you co-founded Coursera so tell us a little bit more about that platform so I founded Coursera as a result of work that I'd been doing at Stanford working on how technology can make education better and more accessible this was a project that I did here a number of my colleagues as well and at some point in the fall of 2011 there was an experiment let's take some of the content that we've been we've been developing within it's within Stanford and put it out there for people to just benefit from and we didn't know what would happen would it be a few thousand people but within a matter of weeks with minimal advertising other than one New York Times article that went viral we had a hundred thousand people in each of those courses and that was a moment in time where you know we looked at this and said can we just go back to writing more papers or is there an incredible opportunity to transform access to education to people all over the world and so I ended up taking a what was supposed to be a teary leave of absence from Stanford to go and co-found Coursera and I thought I'd go back after two years but the but at the end of that two-year period the there was just so much more to be done and so much more impact that we could bring to people all over the world people of both genders people of the different social economic status every single country around the world we I just felt like this was something that I couldn't not do and how did you why did you decide to go from an educational platform to then going into machine learning and biomedicine so I've been doing Coursera for about five years in 2016 and the company was on a great trajectory but it's primarily a Content company and around me machine learning was transforming the world and I wanted to come back and be part of that and when I looked around I saw machine learning being applied to ecommerce and the natural language and to self-driving cars but there really wasn't a lot of impact being made on the life science area and I wanted to be part of making that happen partly because I felt like coming back to our earlier comment that in order to really have that impact you need to have someone who speaks both languages and while there's a new generation of researchers who are bilingual in biology and in machine learning there's still a small group and there very few of those in kind of my age cohort and I thought that I would be able to have a real impact by building and company in the space so it sounds like your background is pretty varied what advice would you give to women who are just starting college now who may be interested in a similar field would you tell them they have to major in math or or do you think that maybe like there are some other majors that may be influential as well I think there's a lot of ways to get into data science math is one of them but there's also statistics or physics and I would say that especially for the field that I'm currently in which is at the intersection of machine learning data science on the one hand and biology and health on the other one can get there from biology or medicine as well but what I think is important is not to shy away from the more mathematically oriented courses in whatever major you're in because that found the is a really strong one there's a lot of people out there who are basically lightweight consumers of data science and they don't really understand how the methods that they're deploying how they work and that limits them in their ability to advance the field and come up with new methods that are better suited perhaps to the problems that they're tackling so I think it's totally fine and in fact there's a lot of value to coming into data science from fields other than a third computer science but I think taking courses in those fields even while you're majoring in whatever field you're interested in is going to make you a much better person who lives at that intersection and how do you think having a technology background has helped you in in founding your companies and has helped you become a successful CEO in companies that are very strongly Rd focused like like in C tro and others having a technical co-founder is absolutely essential because it's fine to have an understanding of whatever the user needs and so on and come from the business side of it and a lot of companies have a business co-founder but not understanding what the technology can actually do is highly limiting because you end up hallucinating oh if we could only do this and yet that would be great but you can't and people end up oftentimes making ridiculous promises about what technology will or will not do because they just don't understand where the land mines sit and and where you're gonna hit real obstacles and in the path so I think it's really important to have a strong technical foundation in these companies and that being said where do you see an teacher in the future and and how do you see it solving say Nash that you talked about in your keynote so we hope that in seat row we'll be a fully integrated drug discovery and development company that is based on a slightly different foundation than a traditional pharma company where they grew up in the old approach of that is very much bespoke scientific analysis of the biology of different diseases and then going after targets or our ways of dealing with the disease that are driven by human intuition where I think we have the opportunity to go today is to build a very data-driven approach that collects massive amounts of data and then let analysis of those data really reveal new hypotheses that might not be the ones that the cord with people's preconceptions of what matters and what doesn't and so hopefully we'll be able to over time create enough data and apply machine learning to address key bottlenecks in the drug discovery development process so we can bring better drugs to people and we can do it faster and hopefully at much lower cost that's great and you also mentioned in your keynote that you think that 2020s is like a digital biology era so tell us more about that so I think if you look if you take a historical perspective on science and think back you realize that there's periods in history where one discipline has made a tremendous amount of progress in a relatively short amount of time because of a new technology or a new way of looking at things in the 1870s that discipline was chemistry was the understanding of the periodic table and that you actually couldn't turn lead into gold in the 1900s that was physics with understanding the connection between matter and energy and between space and time in the 1950s that was computing where silicon chips were suddenly able to perform calculations that up until that point only people have been able to do and then in 1990s there was an interesting bifurcation one was the era of data which is related to computing but also involves elements statistics and optimization of neuroscience and the other one was quantitative biology in which biology moved from a descriptive science of techsan amaizing phenomena to really probing and measuring biology in a very detailed and a high-throughput way using techniques like microarrays that measure the activity of 20,000 genes at once Oh the human genome sequencing of the human genome and many others but these two feels kind of evolved in parallel and what I think is coming now 30 years later is the convergence of those two fields into one field that I like to think of as digital biology where we are able using the tools that have and continue to be developed measure biology in entirely new levels of detail of fidelity of scale we can use the techniques of machine learning and data science to interpret what we're seeing and then use some of the technologies that are also emerging to engineer biology to do things that it otherwise wouldn't do and that will have implications in biomaterials in energy in the environment in agriculture and I think also in human health and it's an incredibly exciting space to be in right now because just so much is happening and the opportunities to make a difference and make the world a better place are just so large that sounds awesome Daphne thank you for your insight and thank you for being on cute thank you I'm so neat agario thanks for watching stay tuned for more great

>>from San Francisco. It's the queue covering our essay conference. 2020. San Francisco Brought to you by Silicon Angle Media's >>Welcome Back Around Here at the Cube. Coverage for our conference. Mosconi, South Floor. Bring you all the action day one of three days of cube coverage where the security game is changing, the big players are making big announcements. The market's changing from on premise to cloud. Then hybrid Multi cloud was seeing that wave coming. A great guest here. Barr, our VP of product strategy and co founder of the Mystery, was acquired by Palo Alto Networks. Worries employed now, Rishi. Thanks for coming on. Thank you. Absolutely happy to be here. So, first of all, great journey for your company. Closed a year ago. Half a 1,000,000,000. Roughly give or take 60. Congratulations. Thank you. Big accomplishments. You guys were taken out right in the growth phase. Now at Palo Alto Networks, which we've been following, you know, very careful. You got a new CMO over there, Jean English? No, we're very well. We're very bullish on Palo Alto. Even though that the on premise transitions happening cloud. You guys are well positioned. How's things going things are going fantastic. We're investing a lot in the next Gen security business across the board, as mentioned Prisma Cloud is big business. And then on the other side, which is what I'm part of the cortex family focused on the Security operations center and the efficiencies That's fantastic and, ah, lot off product innovations, investment and the customer pull from an operations perspective. So very excited. You guys had a big announcement on Monday, and then yesterday was the earnings, which really kind of points to the trend that we're seeing, which is the wave to the cloud, which you're well positioned for this transition going on. I want to get to the news first. Then we get into some of the macro industry questions you guys announced the X ore, which is redefining orchestration. Yes. What is this about? What's this news about? Tell us. >> So this news is about Mr was acquired about a year ago as well. This is taking that Mr Platform and expanding it on, expanding it to include a very core piece, which is Intel management. If you look at a traditional saw, what has happened is soccer teams have had the same dead and over the last few years acquired a sword platform such as a mystery security orchestration, automation and response platform. But the Edge Intel team has always been still separate the threat Intel feeds that came in with separate. With this, we are expanding the power of automation and applying doc to the threat intelligence as well. That is, thread intelligence, current state of the art right now. So the current state of the art of threat intelligence is are the larger organizations typically subscribe to a lot of faith, feeds open source feeds and aggregate them. But the challenge is to aggregate them the sit in a repository and nobody knows what to do with them. So the operationalization of those feeds is completely missing. >> So basically, that is going to have data pile. Corpus is sitting there. No one touches it, and then everyone has to. It's a heavy lift. It's a heavy lift, and nobody knows. Cisco sees the value coming out of it. How do you proactively hunt using those? How do you put them to protecting proactively to explain cortex X, or what is it? And what's the value? So the cortex X or as a platform. There are four core pieces, three off which for the core tenants of the misto since the big one is automation and orchestration. So today we roughly integrate with close to 400 different products security and I t products. Why are the FBI on let customers build these work flows come out of the box with close to 80 or 90 different workloads. The idea of these workloads is being able to connect to one product for the data go to another taken action there Automation, orchestration builds a visual book second s case management and this is very critical, right? I mean, if you look at the process side of security, we have never focused as an industry and the process and the human side of security. So how do you make sure every security alert on the process the case management escalation sl A's are all managed. So that's a second piece off cortex. Third collaboration. One of the core tenants of Mr Waas. We heard from customers that analysts do not talk to each other effectively on when they do. Nobody captures that knowledge. So the misto has an inbuilt boardroom which now Cortex X or has the collaboration war room on that is now available to be able to chat among analysts. But not only that charged with the board take actions. The fourth piece, which is the new expanded platform, is the personal management to be able to now use the power of orchestration, automation collaboration, all for threat intelligence feeds as well. Not only the alerts >> so and so you're adding in the threat. Intelligence feeds, yes. So is that visualize ai on the machine Learning on that? How is that being process in real time? How does that on demand work for that fills. So the biggest piece is applying the automation and intelligence to automatically score that on being able to customize the scoring the customer's needs. Customized confidence score perfect. And once you have the high fidelity indicators automatically go block them as an example. If you get a very high fidelity IOC from FBI that this particular domain is the militias domain, you would want to block that in. Your firewall is executed immediately, and that is not happening today. That is the core, and that's because of the constraint is I don't know the data the way we don't know the data and it's manual. Some human needs to review it. Some human needs to go just not being surfaced, just not. So let's get back into some of the human piece. I love the collaboration piece. One of things that I hear all the time in my cube interviews across all the hundreds of events we go to is the human component you mentioned. Yes, people have burnt out. I mean, like the security guys. I mean, the joke was CIOs have good days once in a while, CSOs don't have any good days, and it's kind of a job board pejorative to that. But that's the reality. Is that it works? Yes. We actually okay, if you have another job. Talking of jokes, we have this. Which is what do you call and overwork security analyst. A security analyst, because every one of them >>is over word. >>So this is a huge thing. So, like the ai and some of the predictive analytics trend Is tourist personalization towards the analyst Exactly. This is a trend that we're seeing. What's your view on this? What? You're absolutely We're seeing that trend which is How do you make sure analyst gets to see the data they're supposed to see at the right time? Right. So there's one aspect is what do you bring up to the analyst? What is relevant and you bring it up at the right time to be able to use it. Respond with that. So that comes in one from an ML perspective and machine learning. And our cortex. XDR suite of products actually does a fantastic job of bringing very rich data to the analyst at the right time. And then the second is, can we help analyst respond to it? Can we take the repetitive work away from them with a playbook approach? And that's what the cortex platform brings to that. I love to riff on some future scenarios kind of. I won't say sci fi, but I got to roll a little bit of a future to me. I think security has to get to like a multi player gaming environment because imagine like a first person shooter game, you know where or a collaborative game where it's fun. Because once you start that collaboration, yes, then you're gonna have some are oi around. I saw that already. Don't waste your time or you get to know people. So sharing has been a big part? Yes. How soon do you think we're gonna get to an environment where I won't say like gaming? But that notion of a headset on I got some data. I know you are your reputation. I think your armor, you're you're certifications. Metaphorically putting. I think way have a lot of these aspects and I think it's a very critical point. You mentioned right one of the things which we call the virtual war room and like sex or I was pointing out the fact that you can have analysts sit in front of a collaboration war room not only charge for the appears but charged with a boat to go take care of. This is equivalent to remember that matrix movie plugging and says, you know how to fly this helicopter data and now I do. That's exactly what it is. I think we need to point move to a point where, no matter what the security tool is what your endpoint is, you should not have to learn every endpoint every time the normalization off, running those commands via the collaboration War Room should be dead. I would say we're starting to see in some of the customers are topics or they're using the collaboration war room to run those commands intractably, I would say, though, there's a big challenge. Security vendors do not do a good job normalizing that data, and that is where we're trying to reach you. First of all, you get the award for bringing up a matrix quote in The Cube interview. So props to that. So you have blue teams. Red teams picked the pill. I mean, people are people picking their teams. You know what's what's going on. How do you see the whole Red Team Blue team thing happening? I think that's a really good stuff happening. In my opinion, John, what's going on is right now so far, if you see if I go back three years our adversaries were are committing. Then we started to see this trend off red teaming automation with beach automation and bunch of companies starting to >>do that >>with Cortex X or on similar products, we're starting to now automate the blue team side of things, which is how do you automatically respond how do you protect yourself? How do you put the response framework back there? I think the next day and I'm starting to see is these things coming together into a unified platform where the blue team and the team are part of the same umbrella. They're sharing the data. They're sharing the information on the threat Intel chair. So I see we are a very, very good part. Of course, the adversities. I'm not gonna sit idle like you said about the Dev ops mindset. Heavens, notion of knowledge coming your way and having sharing packages all baked out for you. She doesn't do the heavy lifting. That's really the problem. The data is a problem. So much demand so much off it. And you don't know what is good and what is not. Great. Great conversation again. The Matrix reference about your journey. You've been an entrepreneur and sold. You had a great exit again. Politics is world class blue chip company in the industry public going through a transition. What's it like from an entrepreneur now to the big company? What's the opportunity is amazing. I think journey has been very quick. One. We saw some crazy growth with the misto on. Even after the acquisition, it's been incredibly fast pace. It's very interesting lot of one of the doctors like, Hey, you must be no resting is like, No, the journey is amazing. I think he s Polito Networks fundamentally believe that. We need to know where it really, really fast to keep the adversaries out on. But that's been the journey. Um, and we have accelerated, in fact, some of our product plans that we hard as a start up on delivering much faster. So the journey has been incredible, and we have been seeing that growth Will they picked you guys write up? There's no vesting interesting going on when you guys were on the uphill on the upslope growth and certainly relevance for Palo Alto. So clearly, you know, you haven't fun. People vested arrest when they checked out, You guys look like you're doing good. So I got to ask you the question that when you started, what was the original mission? Where is it now? I mean this Is there any deviation? What's been the kind? Of course you know, this is very, very relevant questions. It's very interesting. Right after the acquisition, we went and looked at a pitch deck, which we presented overseas in mid 2015. Believe it or not, the mission has not changed, not changing iron. It had the same competent off. How do you make the life off a security person? A security analyst? Easy. It's all the same mission by automating more by applying AI and learning to help them further by letting them collaborate. All the aspects off case management process, collaboration, automation. It's not changed. That's actually very powerful, because if you're on the same mission, of course you're adding more and more capabilities. But we're still on the same path on going on that. So every company's got their own little nuanced. Moore's Law for Intel. What made you guys successful was that the culture of Dev ops? It sounds like you guys had a certain either it was cut in grain. I think I would say, by the way, making things easy. But you got to do it. You got to stay the course. What was that? I think that's a fundamental cultural feature. Yeah, there's one thing really stand by, and I actually tweeted about a few weeks ago, this which is every idea, is as good as good as its execution. So there's two things between really focus on which is customer focused on. We were really, really portable about customer needs to get the product needs to use the product, customer focus and execution. As we heard the customers loud and clear, every small better. And that's what we also did. You guys have this agile mindset as well, absolutely agile mindset and the development that comes with the customer focus because way kind of these micro payments customer wants this like, why do they want this? What is the end goal? Attributed learner. Move on to make a decision making line was on Web services Way debate argue align! Go Then go. And then once you said we see great success story again Startup right out of the gate 2015. Acquire a couple years later, conventions you and your team and looking forward to seeing your next Palo Alto Networks event. Or thanks for coming on. Great insight here on the cube coverage. I'm John Furrier here on the ground floor of our S e commerce on Mosconi getting all the signal extracting it from the noise here on the Cube. Thanks for watching. >>Yeah, yeah,